force and shape estimation using fiber bragg grating ... · i would like to thank dr. abhijit lele...

Force and Shape Estimation using Fiber Bragg Grating

Sensors for Assistance in

Minimally Invasive Diagnostic and Surgical

Procedures

A THE SI S

SUBMITTED FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

IN T HE FA CULTY OF ENGI NEERING

by

Kumar Saurabh

DEPART MENT O F INS TR UMENT ATION AND APP LI ED PHY SI CS

IN DIAN IN STIT UT E O F SCI EN CE

BANGALORE – 560 012

JULY 2017

Declaration

I declare that the thesis entitled “Force and Shape Estimation using Fiber Bragg Grating

Sensors for Assistance in Minimally Invasive Diagnostic and Surgical Procedures”

submitted by me for the Ph.D. degree of the Indian Institute of Science did not form the subject

matter for any thesis submitted by me for any outside Degree and the original work done by

me and incorporated in the thesis is entirely done at the Indian Institute of Science, Bangalore.

Kumar Saurabh

Dedicated to my family and my mentors.

Acknowledgement

Writing this thesis presents me the opportunity to spend some time to think about all the

people who made this journey possible. The External Registration Program at the Indian

Institute of Science provided me the chance to pursue my Ph.D. while still continuing

to work. This allowed me to fulfil my dream which would not have been possible

otherwise.

Throughout this journey, my wife Seema and daughter Surabhi have been extremely

supportive and no words from my side are sufficient to convey my gratitude. Every time

when I was able to conduct experiments till late and night and sleep in the lab leading

to the results I have achieved, they deserved the credit for their patience and

understanding.

The journey started with the head of department of my unit at Robert Bosch Engineering

and Business Solutions Ltd, Dr. Detlef Zerfowski, suggesting that I should pursue a

Ph.D. I had always wanted to do it, but without his support it would never have been

possible to initiate the process. Over the course of the process, I received tremendous

support from Mr. Gaur Dattatreya, Mrs. Pratibha and Mr. Sathyanarayana T.K. in

getting all the approvals.

I have been fortunate to have Prof. Asokan Sundarrajan and Prof. Bharadwaj Amrutur

as my advisers. I have learnt a lot about research, rigor and presentation from them. But

the learning I value most is to be humble and respect everyone I work with. Ph.D. can

be a long journey requiring consistent hard work. Seeing Prof. Asokan and Prof.

Bharadwaj work so hard in the presence of so many challenges has kept me motivated

ii

to push my limits. I will always remember when I sent a paper for review to Prof.

Bharadwaj at 5:00 A.M. in the morning and got back the review comments within an

hour. I will remember this whenever I know that someone else depends on my response.

Working from FBG sensors requires a lot of skills which can be best learnt from an

expert. I am grateful to my seniors Guruprasad and Sridevi who helped me to learn how

to use all the equipment, especially the sensor fabrication setup.

Designing experiments for interdisciplinary research is both fascinating and

challenging. I would like to thank Dr. Srikanth and Dr. Shanthanu from Dept. of

Mechanical Engineering with whom I collaborated to conduct my experiments. Without

their suggestions, designing some of the experimental setups would not have been

possible. I would also like to thank Prof. G.K. Ananthasuresh for his valuable thought

provoking suggestions. I would also like to thank Dr. Ravi Nayar from C.G.S. Hospital,

Bangalore and Dr. Nageshwar Reddy from Asian Institute of Gastroenterology,

Hyderabad for their valuable suggestions which led to some of the experiments. Being

able to do an endoscopy procedure using a pig model would not have been possible

without Dr. Reddy’s help.

I would like to thank Dr. Abhijit Lele and Mr. Sangeeth Nambiar at Bosch Research

and Technology Centre – India for giving the freedom to pursue my research interests

without being entangled in the administrative complications in the office.

I will cherish forever the time spent with my lab-mates Dr.Srinivas, Dr. Srividya, Dr.

Tamilarasan, Dr. Chandasree, Dr. Sharath and Dr. Sikha. Without a lively group, it

would have been difficult to sustain the motivation over such a long time.

I might have missed some names but I would like to thank all the people at Indian

Institute of Science and Bosch who made it possible for me to pursue my research.

iii

Abstract

Fiber Bragg Grating (FBG) sensors have become increasingly popular for

applications in biomedical engineering. Amongst these, use of FBGs in ‘smart

instruments’ for minimally invasive diagnostic and surgical procedures look promising

as FBG sensors provide significant advantages compared to other sensing modalities

in terms of size, electromagnetic immunity, and ability to withstand sterilization

procedures.

Minimally invasive procedures provide lower discomfort to the patients and

faster healing time. Procedures based on use of needles as well as flexible medical

instruments like endoscopes form an important part of these procedures. As a natural

progression, there is a growing trend towards robotic and robot-assisted procedures.

Effective sensing of interactions between the instruments and tissues and the state of the

devices plays an important role in this. This thesis showcases the effectiveness of FBG

sensors in the estimation of forces during device-tissue interactions and the shape of

flexible devices through two applications.

The first application demonstrates the feasibility of estimating needle transitions

through tissues using force estimation at the needle tip. Needles with integrated fiber

Bragg grating sensors have been developed for this purpose and experiments have been

conducted using multi-layered Polydimethylsiloxane (PDMS) phantoms. The design has

been extended to handle temperature induced effects and the experiments have also been

performed using heated chicken tissue.

The second application demonstrates the feasibility of estimating the shape of a

flexible medical instrument like endoscope using strain information from fiber Bragg

grating sensors embedded in a polymer filled tube. This overcomes some of the

iv

constraints of solutions based on sensors bonded on nitinol wires which have been used

earlier. The computation of shape from strain data has been explained.

Widespread use of FBG sensors in such applications is dependent not only on

sensor characteristics but also on the interrogation system. Since the number of sensors

per fiber and the distance between the source and the sensors are small in these

applications, interrogation systems based on linear detector arrays provide a good

option. However, their accuracy depends on the curve fitting method used. For this

purpose, a comparison of accuracies of interrogation systems based on swept tunable

laser and InGaAs linear detector arrays has been performed. The choice and

effectiveness of curve fitting techniques to achieve accuracies similar to tunable laser

based systems have been investigated. The computational feasibility of the algorithm on

embedded hardware has been demonstrated.

1

Contents

Chapter 1: Introduction .............................................................. 13

1.1 Minimally Invasive Procedures ...................................................................... 13

1.2 Robotic and Robot Assisted Procedures ........................................................ 15

1.3 Significance of Force and Shape Estimation and Role of Sensors ................ 16

1.4 Thesis Organization........................................................................................ 17

References ................................................................................................................. 19

Chapter 2: Fiber Bragg Gratings as Sensors for Biomedical

Applications ................................................................................ 35

2.1 Fiber Bragg Gratings: evolution and use as sensors ..................................... 35

2.2 Sensor Fabrication .......................................................................................... 37

2.3 Use in Biomedical Application ...................................................................... 38

2.4 Thesis contributions towards novel smart instrument design using FBG

sensors ....................................................................................................................... 40

References ....................................................................................... 42

Chapter 3: Estimating Needle Transitions through Tissue Layers

.................................................................................................... 51

3.1 Introduction .................................................................................................... 51

3.2 Needle and Sensor Characteristics ................................................................. 55

3.3 Needle Insertion Setup ................................................................................... 58

3.4 Polydimethylsiloxane (PDMS) Phantom ....................................................... 60

3.5 Measurements and Results ............................................................................. 60

3.5.2 Sensitivity Determination ....................................................................... 60

3.5.2 Transition through layers with different stiffness ................................... 61

3.6 Temperature Compensation ........................................................................... 69

2

3.7 Insertion in heated chicken tissue .................................................................. 71

3.8 Summary ....................................................................................................... 74

References ....................................................................................... 75

Chapter 4: Shape Estimation of Flexible Medical Instruments . 81

4.1 Introduction ................................................................................................... 81

4.2 Need for shape estimation and assistance ..................................................... 83

4.3 Shape Estimation Methods ............................................................................ 86

4.4 Shape Estimation from curvature using beam theory ................................... 88

4.5 Strain estimation using FBG sensors bonded on endoscope and on a nitinol

wire 89

4.6 Shape reconstruction from curvature............................................................. 94

4.7 Curvature estimation using FBG sensors embedded in a PDMS filled highly

flexible tube .............................................................................................................. 97

4.8 Summary ..................................................................................................... 103

References ..................................................................................... 104

Chapter 5: Evaluation of FBG interrogation systems based on

InGaAs linear detector arrays and curve fitting methods using

Gaussian approximation ........................................................... 109

5.1 Introduction ................................................................................................. 109

5.2 Measurement Setup and Data Characteristics ............................................. 115

5.2.1 Setup ..................................................................................................... 115

5.2.2 Data Characteristics ............................................................................. 116

5.3 Bragg Wavelength Estimation ..................................................................... 117

5.3.1 Centroid Algorithm .............................................................................. 118

5.3.2 Gaussian Approximation for the FBG spectrum and least square

estimation of Bragg wavelength ......................................................................... 120

5.3.3 Lower computational cost algorithm utilizing Gaussian Approximation

for the FBG spectrum ......................................................................................... 127

5.4 Summary ..................................................................................................... 130

3

References ..................................................................................... 132

Chapter 6: Conclusions and Future Work ................................ 139

Appendix A .............................................................................. 141

FBG sensor fabrication setup .................................................................................. 141

List of Publications ................................................................... 143

5

List of Tables Table 3.1 Inner and outer diameters for common needle gauge sizes .......................... 52

Table 4.1 Flexible endoscope types with typical dimensions ....................................... 82

Table 4.2 Estimated Radius of curvature using strains from FBG sensor bonded on the

endoscope .............................................................................................................. 90

Table 4.3 Results of curvature measurements using nitinol wire ................................. 91

Table 5.1 Wavelength shift difference from tunable laser based system using different

number of points for non-linear least squares estimation using Levenberg

Marquardt algorithm ........................................................................................... 125

Table 5.2. Wavelength shift difference from tunable laser based system using

Levenberg Marquardt algorithm and Centroid Algorithm.................................. 126

Table 5.3 Comparison of execution times and difference in estimated wavelength from

tunable laser based system for the three methods ............................................... 130

7

List of Figures

Figure 2.1 Reflected peak and transmitted spectrum for a fiber Bragg grating ........... 36

Figure 2.2 Schematic diagram of the phase mask method ........................................... 38

Figure 3.1 (a) Bevel tipped and (b) Franseen needle ................................................... 52

Figure 3.2 Stages of needle insertion: (a) tissue deflection (b) rupture (c) movement

inside tissue ........................................................................................................... 53

Figure 3.3 Simulation configuration with one fixed end ............................................. 55

Figure 3.0.4 Load of 0.1N applied at cutting edge ...................................................... 56

Figure 3.5 Fabrication of two FBG sensors close to each other by translating and pre-

stretching the fiber ................................................................................................ 57

Figure 3.6 Needle with FBG sensors bonded ............................................................. 58

Fig 3.7 Schematic diagram of the needle insertion setup ............................................. 59

Figure 3.8 (a) Complete setup including needle translation stage, controller and data

acquisition setup and (b) Needle being inserted in the phantom with a camera

placed to monitor its progress inside the phantom................................................ 59

Figure 3.9 Strain response of the two FBG sensors .................................................... 61

Figure 3.10 Phantoms with five layers with alternating higher and lower stiffness .... 62

Figure 3.11 Stages during needle insertion in the phantom ......................................... 63

Figure 3.12 Response of the sensors during needle insertion ...................................... 64

Figure 3.13 Gradients of the strains and the force at the needle holder base ............... 65

Figure 3.14 Strains for insertion at 3mm/sec ............................................................... 66

Figure 3.15 Gradients for insertion at 3mm/sec ........................................................... 66

Figure 3.16 Strains measured by the two FBG sensors and force measured at the

needle holder base for insertion in phantom 1 at varying needle insertion rates .. 67

8


needle holder base for insertion in phantom 2 at varying needle insertion rates . 68

Figure 3.18 Additional hollow needle with FBG sensor used instead of stylet .......... 69

Figure 3.19 Temperature response of actual FBG sensor and the compensation FBG

sensor ................................................................................................................... 70

Figure 3.20 Setup for insertions in heated chicken tissue ........................................... 71

Figure 3.21 Temperature induced wavelength shift during needle insertion .............. 72

Figure 3.22 (a) Response of the sensor bonded to the 18G needle and that of the

inside needle during insertion in heated tissue and (b) the temperature

compensated response of the sensor bonded to the 18G needle .......................... 73

Figure 3.23 Gradients of strain measured by senor 1 bonded on the 18G needle after

temperature compensation ................................................................................... 74

Figure 4.1 (a) Schematic of Endoscope and (b) holding an endoscope for procedure 81

Figure 4.2 Sketch of endoscope looping during colonoscopy ..................................... 84

Figure 4.3 (a) Sketch showing formation of an ‘n’ loop and (b) pushing the scope

further pushes it against the wall of the colon instead of moving it forward ....... 85

Figure 4.4 Simple image showing an endoscopic training system (Image created as

part of ‘Cyber surgery and remote patient care’ project at Robert Bosch Center

for Cyber Physical Systems, Indian Institute of Science which has led to the

startup Mimyk [10]) ............................................................................................. 86

Figure 4.5 Beam deflection ......................................................................................... 88

Figure 4.6 Setup for curvature estimation using FBG sensor bonded on the endoscope

.............................................................................................................................. 89

Figure 4.7 Experimental setup with FBG sensor bonded to a nitinol wire ................. 91

Figure 4.8 (a) spectrum for unstrained sensor and (b) spectrum for sensor at strain of

5500µstrains during bending ................................................................................ 92

9

Figure 4.9 radius of curvature for bends during endoscopy ........................................ 93

Figure 4.10 Setup to qualitatively evaluate the impact of the nitinol wire put inside the

endoscope on the flexibility of the endoscope. The left end is fixed while the right

end is translated to change the bending radius of the endoscope. A force sensor

on the right side measures the force at the tip of the endoscope. .......................... 93

Figure 4.11 Estimating shape using curvatures measured by a sensing point close to

the tip moving along the structure ........................................................................ 94

Figure 4.12 Reconstruction using approximate circular segments .............................. 96

Figure 4.13 Cross section of the shape sensing tube .................................................... 97

Figure 4.14 Calibration setup – the inside tube is the sensing tube with FBG sensors

............................................................................................................................... 98

Figure 4.15 Examples of wavelength shifts for varying radii of curvature and

rotational position ................................................................................................. 99

Figure 4.16 Plot showing wavelength shifts for different radii of curvature ............. 100

Figure 4.17 An artificial neural network to perform regression for estimation of radius

of curvature. The shown ANN shows the structure and weights after training

based on data presented in figure 4.16 ................................................................ 101

Figure 4.18 Error plot for tests using the trained neural network .............................. 102

Figure 4.19 Reconstruction of a sample curve using measured wavelength shifts and

estimated radii of curvature using the ANN ....................................................... 103

Figure 5.1 Schematic diagram of interrogation system using edge filter [3] ............. 110

Figure 5.2 (a) Schematic of matched FBGs based interrogation and (b) simultaneous

multiple sensor interrogation [7] [8] ................................................................... 111

Figure 5.3 Schematic diagram of interrogation system using tunable fabry-perot filter

[15] [16] .............................................................................................................. 112

10

Figure 5.4 Schematic diagram of an interrogation system with InGaAs linear detector

array [26] ............................................................................................................ 114

Figure 5.5 Setup to generate data samples ................................................................ 116

Figure 5.6 Spectrum obtained from the linear detector array ................................... 117

Figure 5.7 Errors in peak wavelength estimation using the centroid method compared

to measurement using the interrogation system based on swept tunable laser .. 119

Figure 5.8 Plots showing the points around the peak. Each measurement corresponds

to a shift in wavelength due to applied strain. The original and the shifted points

and the estimated peak using centroid algorithm are shown for three different

sensors on the fiber............................................................................................. 120

Figure 5.9 Estimated Bragg wavelengths using different number of pixels around the

peak .................................................................................................................... 122

Figure 5.10 Wavelength shift measured using different number of pixels and

compared to measurement with swept tunable lased based interrogation system.

............................................................................................................................ 123

Figure 5.11 Spectrum for sensor 1 and sensor 2 before and after wavelength shift due

to applied strain .................................................................................................. 124

Figure 5.12 Difference in wavelength shift estimation using Levenberg-Marquardt

algorithm and Gaussian spectrum approximation compared to measurement using

swept tunable laser based interrogation system ................................................. 126

11

List of Symbols

λb Bragg wavelength

Δλ Bragg wavelength shift

Λ grating pitch

Λpm phase mask period

pe photo-elastic coefficient of the fiber

αn Thermo-optic coefficient of the fiber

αΛ Thermal expansion coefficient of the fiber

neff effective refractive index

l length

Δl change in length

ε strain

T temperature

ΔT change in temperature

R Radius of curvature

T unit tangent

N unit normal

B unit bi-normal

κ curvature

Ii intensity at ith pixel

λi wavelength corresponding to ith pixel

Chapter 1 Introduction

13

Chapter 1

Introduction

1.1 Minimally Invasive Procedures

The origins of minimally invasive diagnosis can be traced back to a rectal speculum

described by Hippocrates (460-375 B.C.). However, the credit for the first instrument

proposed for being used for diagnosis through a natural orifice goes to Bozzini’s

‘Lichtleiter’ (light conductor) [1]. Even though the idea created controversy and drew

criticism [2], it has laid the foundation for a field which has evolved through the last

two centuries and has significantly impacted the practice of medicine. Several

innovations like the development of flexible fiber-optic endoscope [3], the adoption of

needle biopsies [4], the use of brachytherapy [5] and the introduction of the

laparoscopic cholecystectomy (surgical removal of gall bladder) by Mouret [6], have

been landmarks in the evolution of minimally invasive diagnostic, therapeutic and

surgical procedures. The term ‘minimally invasive surgery’ can be attributed to John

Wicham [7]. His prediction that such techniques will be developed for several types of

surgery has come true [8] [9] [10] [11] [12] [13] [14]. It has even led to extreme

experiments like transatlantic robot assisted tele-surgery [15] and even tele-robotic

procedures in low bandwidth settings [16].

Minimally invasive procedures cover a wide range of diagnostic, therapeutic and

surgical procedures. However, an approach to categorization can be made based on the

type of instrument used which are:


14

i) flexible endoscopes (e.g. through a natural orifice as in gastrointestinal

endoscopy, colonoscopy and bronchoscopy)

ii) laparoscope (e.g. in cholecystectomy)

iii) needles (e.g. in biopsies, ablation, brachytherapy and neurosurgery)

iv) catheters (e.g. in angioplasty and stenting)

Minimally invasive procedures provide several advantages [17] [18] [19] [20] [21] [22]

[23] such as:

Short hospital stay and quick recovery time

Less trauma

Lass scarring

Lower risk of complications

Lower risk of infection

Even though there are several advantages for the patients, minimally invasive

procedures present new challenges and constraints for the doctors. These can be split

into two parts – 1) challenges in performing the procedures, and 2) challenges in

acquiring new skills through training.

The constraints and challenges in performing the procedures are related to [24] [25]

[26]:

constrained operating environment

lack of depth perception

loss of tool orientation and pose

loss of sense of touch and haptic feedback

high effort in maneuvering and risk of perforation


15

The constraints and challenges in skill acquisition are related to [27] [28] [29] [30]:

high initial learning curve

availability of realistic training systems

transfer of skills acquired for one task to other tasks

1.2 Robotic and Robot Assisted Procedures

The motivation for minimally invasive procedures has been to reduce patient discomfort

and the risks associated with the procedure. Over the years, some minimally invasive

procedures in cancer management, cardiac treatments and neurosurgery have become

routine. However, some advanced techniques like Natural Orifice Transluminal Surgery

(NOTES) [31] are still evolving [32] [33] and the efficacy, learning curve [34], role of

new devices [35] [36] and patient benefits are still under discussion [37]. The same is

applicable for robotic and robot assisted procedures [38] [39] [40] [41] with very few

systems like the da Vinci® system available commercially. In such systems, the

motivation has been to help the doctors perform the procedures better by creating

devices with higher degrees of freedom, improved visualization and guidance, and

reducing effects of hand tremors. In its current manifestation, surgical robotics is

restricted to systems which are very constrained in their applications. However, the

consensus document [42] on robotic surgery approved by the Board of Governors of the

Society of American Gastrointestinal and Endoscopic Surgeons (SAGES) suggests that

research in the area should focus on future smart instruments that can provide ‘smart

sensing’ capabilities to provide surgeons with information like blood flow, oxygenation

and tumor margins. It also suggests that future systems can use the knowledge of spatial

coordinates along with preoperative and intra operative imaging providing overlays to

enable identifying danger zones or organ boundaries. Another expectation from future

systems is to overcome the rigidity and size constraints of current systems through the


16

use of more flexible components. The effects of robotic assistance on learning curves

also have to be addressed [43].

1.3 Significance of Force and Shape Estimation and

Role of Sensors

Minimally invasive procedures create a constrained operating environment leading to

loss of tool orientation and pose and lack of haptic feedback. Both these have attracted

significant research, both in terms of creating improved sensing-enabled devices acting

as ‘smart instruments’ as well as improved visualizations.

Estimating forces as part of tool-tissue interactions is an important aspect of these smart

instruments enabling easier and improved tool manipulation and haptic feedback. These

advantages get amplified when forces cannot be directly transmitted easily to the

operating surgeon’s hand due to device configuration or conversely when the forces

applied at the distal end have to be regulated or the effects of tremors have to be

controlled. Several studies [44] [45] [46] [47] [48] [49] [50] [51] have evaluated the

need and value of estimating these forces and providing feedback. These have led to the

development of novel instruments and end effectors with force measurement

capabilities, such as sensing-enabled endoscopic forceps [52] [53] and graspers [54]

[55] [56] [57], indentation [58] [59] and palpation probes [60] [61], scissor blades [62],

catheters [63] [64] [65], instruments for NOTES [66], neurosurgery [67], and

microsurgery [68] [69]. The choice of sensors plays an important role in development

of these ‘smart instruments’. Several sensor types including Micro Electro Mechanical

Sensors (MEMS) [70] [71], ultrasonic sensors [72] [73], piezoelectric sensors [74] [75]

and optical sensors [76] [77] [78] [79] [80] have been used.

Shape Estimation plays an important role in determining the location and pose of the

end effector as well as the entire instrument. This has higher significance for flexible


17

instruments where loss of tool orientation during the procedure is more common [37]

or where dynamic shape estimation is important for guidance to effectively reach certain

locations [81]. Similar to force estimation, shape estimation has attracted significant

research. Even though radiological methods like fluoroscopy [82] can be used, it has its

own challenges. There has been effort to reduce the radiation exposure by reducing the

number of X-Ray images required [83]. However, this is still not very suitable for

continuous monitoring. Introduction of robotic and robot assisted procedures has

contributed to additional interest in continuous shape and pose estimation. It has

gathered special significance with recent advances through use of continuum

reconfigurable robots [84] and concentric tube robots [85] [86]. Issues with radiological

procedures have led to development of non-radiological methods including the use of

electromagnetic trackers [87] [88] [89], MEMS [90] [91], optical sensors [92] [93] and

fusion of multiple sensing modalities [94] [95].

1.4 Thesis Organization

In recent years, Fiber Bragg Grating (FBG) sensors have become popular for

applications that require small sensor size, multiplexing and electromagnetic

immunity. This thesis explores the use of FBG sensors in force and shape

estimation for applications in minimally invasive procedures.

Chapter two of the thesis presents an overview of sensing capabilities of Fiber Bragg

Gratings and their use in biomedical applications including medical robotics. The sensor

fabrication method for sensors used in this work is briefly described.

Needles are an important part of several diagnostic, radiological and neurosurgical

procedures. Chapter three presents a method to estimate transition of hollow needles

through tissue layers using force estimation at needle tip using fiber Bragg grating

sensors.


18

Flexible instruments are an integral part of minimally invasive procedures and more and

more procedures are being done using flexible instruments like endoscopes. A method

to estimate shape for highly flexible instruments is presented in chapter four.

Widespread use of fiber Bragg grating sensor for medical robotic applications depend

not just on sensor characteristics but also on cost effective interrogation systems.

Chapter five provides an analysis of accuracy of interrogation systems based on InGaAs

linear detector arrays and computationally efficient curve fitting algorithm implemented

on embedded hardware for these applications that can enable cost effective interrogation

systems which can be integrated in existing medical setups.

The publications based on the work presented in chapters three, four and five have been

listed at the end of the thesis.


19

References

[1] P. Bozzini, Lichtleiter, eine Erfindung zur Anschauung innerer Theile und

Krankheiten nebst der Abbildung, 1806.

[2] J. Shah, "Endoscopy through the ages," BJU international, vol. 89, pp. 645-652,

2002.

[3] B. I. Hirschowitz, "A personal history of the fiberscope," Gastroenterology, vol.

76, pp. 864-869, 1979.

[4] G. Vlastos and H. M. Verkooijen, "Minimally invasive approaches for diagnosis

and treatment of early-stage breast cancer," The oncologist, vol. 12, pp. 1-10,

2007.

[5] H. Ragde, G. L. Grado, B. Nadir and A.-A. Elgamal, "Modern prostate

brachytherapy," CA: a cancer journal for clinicians, vol. 50, pp. 380-393,

2000.

[6] A. Cuschieri, F. Dubois, J. Mouiel, P. Mouret, H. Becker, G. Buess, M. Trede and

H. Troidl, "The European experience with laparoscopic cholecystectomy,"

The American journal of surgery, vol. 161, pp. 385-387, 1991.

[7] J. E. Wickham, "The new surgery.," British medical journal (Clinical research

ed.), vol. 295, p. 1581, 1987.


20

[8] K. a. W. E. Jr, "The evolution of laparoscopy and the revolution in surgery in the

decade of the 1990s," Jsls, vol. 12, pp. 351-357, 2008.

[9] M. B. Wallace, J. M. S. Pascual, M. Raimondo, T. A. Woodward, B. L. McComb,

J. E. Crook, M. M. Johnson, M. A. Al-Haddad, S. A. Gross, S. Pungpapong

and others, "Minimally invasive endoscopic staging of suspected lung

cancer," Jama, vol. 299, pp. 540-546, 2008.

[10] A. G. Harrell and B. T. Heniford, "Minimally invasive abdominal surgery: lux et

veritas past, present, and future," The American journal of surgery, vol. 190,

pp. 239-243, 2005.

[11] S. Bhatnagar and Y. K. Sarin, "Scope and limitations of minimal invasive surgery

in practice of pediatric surgical oncology," Indian journal of medical and

paediatric oncology: official journal of Indian Society of Medical \&

Paediatric Oncology, vol. 31, p. 137, 2010.

[12] G. C. Vitale, B. R. Davis and T. C. Tran, "The advancing art and science of

endoscopy," The American journal of surgery, vol. 190, pp. 228-233, 2005.

[13] E. A. Te Velde, N. M. A. Bax, S. H. A. J. Tytgat, J. R. De Jong, D. V. Travassos,

W. L. M. Kramer and D. C. van der Zee, "Minimally invasive pediatric

surgery: increasing implementation in daily practice and resident’s training,"

Surgical endoscopy, vol. 22, pp. 163-166, 2008.


21

[14] C. Tsui, R. Klein and M. Garabrant, "Minimally invasive surgery: national trends

in adoption and future directions for hospital strategy," Surgical endoscopy,

vol. 27, pp. 2253-2257, 2013.

[15] J. Marescaux, J. Leroy, M. Gagner, F. Rubino, D. Mutter, M. Vix, S. E. Butner

and M. K. Smith, "Transatlantic robot-assisted telesurgery," Nature, vol.

413, pp. 379-380, 2001.

[16] J. C. Rosser, R. L. Bell, B. Harnett, E. Rodas, M. Murayama and R. Merrell, "Use

of mobile low-bandwith telemedical techniques for extreme telemedicine

applications," Journal of the American College of Surgeons, vol. 189, pp.

397-404, 1999.

[17] J. C. Hu, X. Gu, S. R. Lipsitz, M. J. Barry, A. V. D’Amico, A. C. Weinberg and

N. L. Keating, "Comparative effectiveness of minimally invasive vs open

radical prostatectomy," Jama, vol. 302, pp. 1557-1564, 2009.

[18] K. Nagpal, K. Ahmed, A. Vats, D. Yakoub, D. James, H. Ashrafian, A. Darzi, K.

Moorthy and T. Athanasiou, "Is minimally invasive surgery beneficial in the

management of esophageal cancer? A meta-analysis," Surgical endoscopy,

vol. 24, pp. 1621-1629, 2010.

[19] S. S. A. Y. Biere, I. van Berge Henegouwen and Mark, K. W. Maas, L. Bonavina,

C. Rosman, J. R. Garcia, S. S. Gisbertz, J. H. G. Klinkenbijl, M. W.

Hollmann, S. M. de Lange and Elly and others, "Minimally invasive versus

open oesophagectomy for patients with oesophageal cancer: a multicentre,


22

open-label, randomised controlled trial," The Lancet, vol. 379, pp. 1887-

1892, 2012.

[20] L. H. Cohn, D. H. Adams, G. S. Couper, D. P. Bichell, D. M. Rosborough, S. P.

Sears and S. F. Aranki, "Minimally invasive cardiac valve surgery improves

patient satisfaction while reducing costs of cardiac valve replacement and

repair.," Annals of surgery, vol. 226, p. 421, 1997.

[21] S. H. Heywang-Köbrunner, U. Schaumlöffel, P. Viehweg, H. Höfer, J. Buchmann

and D. Lampe, "Minimally invasive stereotaxic vacuum core breast biopsy,"

European radiology, vol. 8, pp. 377-385, 1998.

[22] G. Sgourakis, I. Gockel, A. Radtke, T. J. Musholt, S. Timm, A. Rink, A. Tsiamis,

C. Karaliotas and H. Lang, "Minimally invasive versus open esophagectomy:

meta-analysis of outcomes," Digestive diseases and sciences, vol. 55, pp.

3031-3040, 2010.

[23] D. Nuss, R. E. Kelly, D. P. Croitoru and M. E. Katz, "A 10-year review of a

minimally invasive technique for the correction of pectus excavatum,"

Journal of pediatric surgery, vol. 33, pp. 545-552, 1998.

[24] N. Tanigawa, "Advantages and Problems with Endoscopic Surgery," JMAJ, vol.

52, pp. 330-334, 2009.

[25] C. G. L. Cao and P. Milgram, "Disorientation in minimal access surgery: A case

study," in Proceedings of the Human Factors and Ergonomics Society

Annual Meeting, 2000.


23

[26] M. J. Mack, "Minimally invasive and robotic surgery," Jama, vol. 285, pp. 568-

572, 2001.

[27] A. E. Park and T. H. Lee, "Evolution of minimally invasive surgery and its impact

on surgical residency training," in Minimally Invasive Surgical Oncology,

Springer, 2011, pp. 11-22.

[28] K. E. Roberts, R. L. Bell and A. J. Duffy, "Evolution of surgical skills training,"

World Journal of Gastroenterology, vol. 12, p. 3219, 2006.

[29] C. Richards, J. Rosen, B. Hannaford, C. Pellegrini and M. Sinanan, "Skills

evaluation in minimally invasive surgery using force/torque signatures,"


[30] I. A. Mouzas, "Stimulation through simulation. The use of simulators in

gastrointestinal endoscopy training," ANNALS OF

GASTROENTEROLOGY., vol. 17, pp. 23-25, 2004.

[31] T. H. Baron, "Natural orifice transluminal endoscopic surgery," British journal of

surgery, vol. 94, pp. 1-2, 2007.

[32] S. A. Giday, S. V. Kantsevoy and A. N. Kalloo, "Principle and history of natural

orifice translumenal endoscopic surgery (NOTES)," Minimally Invasive

Therapy & Allied Technologies, vol. 15, pp. 373-377, 2006.

[33] A. A. Gumbs, D. Fowler, L. Milone, J. C. Evanko, A. O. Ude, P. Stevens and M.

Bessler, "Transvaginal natural orifice translumenal endoscopic surgery


24

cholecystectomy: early evolution of the technique," Annals of surgery, vol.

249, pp. 908-912, 2009.

[34] L. Freeman, E. Y. Rahmani, R. C. F. Burgess, M. Al-Haddad, D. J. Selzer, S.

Sherman and P. Constable, "Evaluation of the learning curve for natural

orifice transluminal endoscopic surgery: bilateral ovariectomy in dogs,"

Veterinary Surgery, vol. 40, pp. 140-150, 2011.

[35] S. J. Bardaro and L. Swanström, "Development of advanced endoscopes for

natural orifice transluminal endoscopic surgery (NOTES)," Minimally

Invasive Therapy & Allied Technologies, vol. 15, pp. 378-383, 2006.

[36] D. J. Desilets, T. J. Mader, J. R. Romanelli and D. B. Earle, "Gastric transmural

pressure measurements in vivo: implications for natural orifice transluminal

endoscopic surgery (NOTES)," Gastrointestinal endoscopy, vol. 71, pp. 583-

588, 2010.

[37] M. H. Sodergren, J. Clark, T. Athanasiou, J. Teare, G.-Z. Yang and A. Darzi,

"Natural orifice translumenal endoscopic surgery: critical appraisal of

applications in clinical practice," Surgical endoscopy, vol. 23, pp. 680-687,

2009.

[38] S. Maeso, M. Reza, J. A. Mayol, J. A. Blasco, M. Guerra, E. Andradas and M. N.

Plana, Efficacy of the Da Vinci surgical system in abdominal surgery

compared with that of laparoscopy: a systematic review and meta-analysis,

LWW, 2010.


25

[39] A. R. Lanfranco, A. E. Castellanos, J. P. Desai and W. C. Meyers, "Robotic

surgery: a current perspective," Annals of surgery, vol. 239, pp. 14-21, 2004.

[40] A. Tan, H. Ashrafian, A. J. Scott, S. E. Mason, L. Harling, T. Athanasiou and A.

Darzi, "Robotic surgery: disruptive innovation or unfulfilled promise? A

systematic review and meta-analysis of the first 30 years," Surgical

endoscopy, vol. 30, pp. 4330-4352, 2016.

[41] J. Marescaux, M. K. Smith, D. Fölscher, F. Jamali, B. Malassagne and J. Leroy,

"Telerobotic laparoscopic cholecystectomy: initial clinical experience with

25 patients," Annals of surgery, vol. 234, pp. 1-7, 2001.

[42] D. M. Herron and M. Marohn, "A consensus document on robotic surgery,"


[43] S. M. Prasad, H. S. Maniar, N. J. Soper, R. J. Damiano and M. E. Klingensmith,

"The effect of robotic assistance on learning curves for basic laparoscopic

skills," The American journal of surgery, vol. 183, pp. 702-707, 2002.

[44] G. Tholey, J. P. Desai and A. E. Castellanos, "Force feedback plays a significant

role in minimally invasive surgery: results and analysis," Annals of surgery,

vol. 241, pp. 102-109, 2005.

[45] N. Enayati, E. De Momi and G. Ferrigno, "Haptics in robot-assisted surgery:

challenges and benefits," IEEE reviews in biomedical engineering, vol. 9, pp.

49-65, 2016.


26

[46] O. A. J. Van der Meijden and M. P. Schijven, "The value of haptic feedback in

conventional and robot-assisted minimal invasive surgery and virtual reality

training: a current review," Surgical endoscopy, vol. 23, pp. 1180-1190,

2009.

[47] M. I. Tiwana, S. J. Redmond and N. H. Lovell, "A review of tactile sensing

technologies with applications in biomedical engineering," Sensors and

Actuators A: physical, vol. 179, pp. 17-31, 2012.

[48] A. M. Okamura, "Haptic feedback in robot-assisted minimally invasive surgery,"

Current opinion in urology, vol. 19, p. 102, 2009.

[49] M. E. H. Eltaib and J. R. Hewit, "Tactile sensing technology for minimal access

surgery----a review," Mechatronics, vol. 13, pp. 1163-1177, 2003.

[50] P. Puangmali, K. Althoefer, L. D. Seneviratne, D. Murphy and P. Dasgupta,

"State-of-the-art in force and tactile sensing for minimally invasive surgery,"

IEEE Sensors Journal, vol. 8, pp. 371-381, 2008.

[51] F. Perna, E. K. Heist, S. B. Danik, C. D. Barrett, J. N. Ruskin and M. Mansour,

"Assessment of catheter tip contact force resulting in cardiac perforation in

swine atria using force sensing technologyclinical perspective," Circulation:

Arrhythmia and Electrophysiology, vol. 4, pp. 218-224, 2011.

[52] K. J. Rebello, "Applications of MEMS in surgery," Proceedings of the IEEE, vol.

92, pp. 43-55, 2004.


27

[53] U. Kim, D.-H. Lee, W. J. Yoon, B. Hannaford and H. R. Choi, "Force sensor

integrated surgical forceps for minimally invasive robotic surgery," IEEE

Transactions on Robotics, vol. 31, pp. 1214-1224, 2015.

[54] J. Dargahi and S. Najarian, "An endoscopic force-position sensor grasper with

minimum sensors," Canadian Journal of Electrical and Computer

Engineering, vol. 28, pp. 155-161, 2003.

[55] J. B. Gafford, S. B. Kesner, R. J. Wood and C. J. Walsh, "Force-sensing surgical

grasper enabled by pop-up book MEMS," in Intelligent Robots and Systems

(IROS), 2013 IEEE/RSJ International Conference on, 2013.

[56] M. Salerno, K. Zhang, A. Menciassi and J. S. Dai, "A novel 4-DOF origami

grasper with an SMA-actuation system for minimally invasive surgery,"

IEEE Transactions on Robotics, vol. 32, pp. 484-498, 2016.

[57] J. Rosen, B. Hannaford, M. P. MacFarlane and M. N. Sinanan, "Force controlled

and teleoperated endoscopic grasper for minimally invasive surgery-

experimental performance evaluation," IEEE Transactions on Biomedical

Engineering, vol. 46, pp. 1212-1221, 1999.

[58] E. Samur, M. Sedef, C. Basdogan, L. Avtan and O. Duzgun, "A robotic indenter

for minimally invasive measurement and characterization of soft tissue

response," Medical Image Analysis, vol. 11, pp. 361-373, 2007.


28

[59] H. Liu, J. Li, X. Song, L. D. Seneviratne and K. Althoefer, "Rolling indentation

probe for tissue abnormality identification during minimally invasive

surgery," IEEE Transactions on Robotics, vol. 27, pp. 450-460, 2011.

[60] S. McKinley, A. Garg, S. Sen, R. Kapadia, A. Murali, K. Nichols, S. Lim, S. Patil,

P. Abbeel, A. M. Okamura and others, "A single-use haptic palpation probe

for locating subcutaneous blood vessels in robot-assisted minimally invasive

surgery," in Automation Science and Engineering (CASE), 2015 IEEE

International Conference on, 2015.

[61] J. Konstantinova, A. Jiang, K. Althoefer, P. Dasgupta and T. Nanayakkara,

"Implementation of tactile sensing for palpation in robot-assisted minimally

invasive surgery: A review," IEEE Sensors Journal, vol. 14, pp. 2490-2501,

2014.

[62] D. J. Callaghan, G. Rajan, M. M. McGrath, E. Coyle, Y. Semenova and G. Farrell,

"Investigation and experimental measurement of scissor blade cutting forces

using fiber Bragg grating sensors," Smart Materials and Structures, vol. 20,

p. 105004, 2011.

[63] A. Natale, V. Y. Reddy, G. Monir, D. J. Wilber, B. D. Lindsay, H. T. McElderry,

C. Kantipudi, M. C. Mansour, D. P. Melby, D. L. Packer and others,

"Paroxysmal AF catheter ablation with a contact force sensing catheter:

results of the prospective, multicenter SMART-AF trial," Journal of the

American College of Cardiology, vol. 64, pp. 647-656, 2014.

[64] K.-H. Kuck, V. Y. Reddy, B. Schmidt, A. Natale, P. Neuzil, N. Saoudi, J.

Kautzner, C. Herrera, G. Hindricks, P. Ja\̈is and others, "A novel


29

radiofrequency ablation catheter using contact force sensing: Toccata study,"

Heart rhythm, vol. 9, pp. 18-23, 2012.

[65] S. Haldar, J. W. E. Jarman, S. Panikker, D. G. Jones, T. Salukhe, D. Gupta, G.

Wynn, W. Hussain, V. Markides and T. Wong, "Contact force sensing

technology identifies sites of inadequate contact and reduces acute

pulmonary vein reconnection: a prospective case control study,"

International journal of cardiology, vol. 168, pp. 1160-1166, 2013.

[66] A. L. Trejos, S. Jayaraman, R. V. Patel, M. D. Naish and C. M. Schlachta, "Force

sensing in natural orifice transluminal endoscopic surgery," Surgical

endoscopy, vol. 25, pp. 186-192, 2011.

[67] L. S. Gan, K. Zareinia, S. Lama, Y. Maddahi, F. W. Yang and G. R. Sutherland,

"Quantification of forces during a neurosurgical procedure: A pilot study,"

World neurosurgery, vol. 84, pp. 537-548, 2015.

[68] A. Menciassi, A. Eisinberg, M. C. Carrozza and P. Dario, "Force sensing

microinstrument for measuring tissue properties and pulse in microsurgery,"

IEEE/ASME transactions on mechatronics, vol. 8, pp. 10-17, 2003.

[69] P. J. Berkelman, L. L. Whitcomb, R. H. Taylor and P. Jensen, "A miniature

microsurgical instrument tip force sensor for enhanced force feedback during

robot-assisted manipulation," IEEE Transactions on Robotics and

Automation, vol. 19, pp. 917-921, 2003.

[70] W. Wang, Y. Zhao and Q. Lin, "An integrated MEMS tactile tri-axial micro-force

probe sensor for Minimally Invasive Surgery," in Nano/Molecular Medicine


30

and Engineering (NANOMED), 2009 IEEE International Conference on,

2009.

[71] F. Tendick, S. S. Sastry, R. S. Fearing and M. Cohn, "Applications of

micromechatronics in minimally invasive surgery," IEEE/ASME

transactions on mechatronics, vol. 3, pp. 34-42, 1998.

[72] J. Chen, X. Cheng, C.-C. Chen, P.-C. Li, J.-H. Liu and Y.-T. Cheng, "A capacitive

micromachined ultrasonic transducer array for minimally invasive medical

diagnosis," Journal of Microelectromechanical Systems, vol. 17, pp. 599-

610, 2008.

[73] F. Tatar, J. R. Mollinger, R. C. Den Dulk, W. A. Van Duyl, J. F. L. Goosen and

A. Bossche, "Ultrasonic sensor system for measuring position and

orientation of laproscopic instruments in minimal invasive surgery," in

Microtechnologies in Medicine & Biology 2nd Annual International IEEE-

EMB Special Topic Conference on, 2002.

[74] C. Li, P.-M. Wu, S. Lee, A. Gorton, M. J. Schulz and C. H. Ahn, "Flexible dome

and bump shape piezoelectric tactile sensors using PVDF-TrFE copolymer,"

Journal of Microelectromechanical Systems, vol. 17, pp. 334-341, 2008.

[75] J. Dargahi, "A piezoelectric tactile sensor with three sensing elements for robotic,

endoscopic and prosthetic applications," Sensors and Actuators A: Physical,

vol. 80, pp. 23-30, 2000.


31

[76] P. Puangmali, H. Liu, K. Althoefer and L. D. Seneviratne, "Optical fiber sensor

for soft tissue investigation during minimally invasive surgery," in Robotics

and Automation, 2008. ICRA 2008. IEEE International Conference on, 2008.

[77] H. Song, K. Kim and J. Lee, "Development of optical fiber Bragg grating force-

reflection sensor system of medical application for safe minimally invasive

robotic surgery," Review of Scientific Instruments, vol. 82, p. 074301, 2011.

[78] J. Peirs, J. Clijnen, D. Reynaerts, H. Van Brussel, P. Herijgers, B. Corteville and

S. Boone, "A micro optical force sensor for force feedback during minimally

invasive robotic surgery," Sensors and Actuators A: Physical, vol. 115, pp.

447-455, 2004.

[79] I. Iordachita, Z. Sun, M. Balicki, J. U. Kang, S. J. Phee, J. Handa, P. Gehlbach and

R. Taylor, "A sub-millimetric, 0.25 mN resolution fully integrated fiber-optic

force-sensing tool for retinal microsurgery," International journal of

computer assisted radiology and surgery, vol. 4, pp. 383-390, 2009.

[80] X. He, M. A. Balicki, J. U. Kang, P. L. Gehlbach, J. T. Handa, R. H. Taylor and

I. I. Iordachita, "Force sensing micro-forceps with integrated fiber bragg

grating for vitreoretinal surgery," in SPIE BiOS, 2012.

[81] C. G. L. Cao, "Guiding navigation in colonoscopy," Surgical endoscopy, vol. 21,

pp. 480-484, 2007.


32

[82] D. G. Adler, "The role of fluoroscopy in the endoscopic management of luminal

gastrointestinal disorders," Techniques in Gastrointestinal Endoscopy, vol.

9, pp. 189-194, 2007.

[83] E. J. Lobaton, J. Fu, L. G. Torres and R. Alterovitz, "Continuous shape estimation

of continuum robots using X-ray images," in Robotics and Automation

(ICRA), 2013 IEEE International Conference on, 2013.

[84] P. L. Anderson, A. Mahoney and I. I. I. a. J. R. Webster, "Continuum

Reconfigurable Parallel Robots for Surgery: Shape Sensing and State

Estimation with Uncertainty," IEEE Robotics and Automation Letters, 2017.

[85] H. B. Gilbert, D. C. Rucker and I. I. I. a. R. J. Webster, "Concentric tube robots:

The state of the art and future directions," in Robotics Research, Springer,

2016, pp. 253-269.

[86] A. W. Mahoney, T. L. Bruns, P. J. Swaney and R. J. Webster, "On the inseparable

nature of sensor selection, sensor placement, and state estimation for

continuum robots or “where to put your sensors and how to use them”," in

Robotics and Automation (ICRA), 2016 IEEE International Conference on,

2016.

[87] H. Sadjadi, K. Hashtrudi-Zaad and G. Fichtinger, "Fusion of electromagnetic

trackers to improve needle deflection estimation: simulation study," IEEE

Transactions on Biomedical Engineering, vol. 60, pp. 2706-2715, 2013.


33

[88] X. Luo and K. Mori, "Robust endoscope motion estimation via an animated

particle filter for electromagnetically navigated endoscopy," IEEE


[89] F. N. Mefleh, G. H. Baker and D. M. Kwartowitz, "Heuristic estimation of

electromagnetically tracked catheter shape for image-guided vascular

procedures," in SPIE Medical Imaging, 2014.

[90] Z. Zhang, J. Shang, C. Seneci and G.-Z. Yang, "Snake robot shape sensing using

micro-inertial sensors," in Intelligent Robots and Systems (IROS), 2013

IEEE/RSJ International Conference on, 2013.

[91] T. N. Kulatunga, R. A. G. P. Ranasinghe, R. A. C. Ranathunga, R. A. C. H.

Ratnayake and N. D. Nanayakkara, "Real time endoscope trajectory tracking

in the 3D space using MEMS sensors," in Industrial and Information Systems

(ICIIS), 2013 8th IEEE International Conference on, 2013.

[92] R. J. Roesthuis, S. Janssen and S. Misra, "On using an array of fiber bragg grating

sensors for closed-loop control of flexible minimally invasive surgical

instruments," in Intelligent Robots and Systems (IROS), 2013 IEEE/RSJ

International Conference on, 2013.

[93] K. K. C. Lee, A. Mariampillai, M. Haque, B. A. Standish, V. X. D. Yang and P.

R. Herman, "Temperature-compensated fiber-optic 3D shape sensor based

on femtosecond laser direct-written Bragg grating waveguides," Optics

express, vol. 21, pp. 24076-24086, 2013.


34

[94] C. Shi, C. Tercero, X. Wu, S. Ikeda, K. Komori, K. Yamamoto, F. Arai and T.

Fukuda, "Real-time in vitro intravascular reconstruction and navigation for

endovascular aortic stent grafting," The International Journal of Medical

Robotics and Computer Assisted Surgery, 2016.

[95] H. Ren and P. Kazanzides, "Investigation of attitude tracking using an integrated

inertial and magnetic navigation system for hand-held surgical instruments,"

IEEE/ASME Transactions on Mechatronics, vol. 17, pp. 210-217, 2012.

Chapter 2 Fiber Bragg Gratings as Sensors for

Biomedical Applications

35

Chapter 2

Fiber Bragg Gratings as Sensors for


2.1 Fiber Bragg Gratings: evolution and use as sensors

The discovery of photosensitivity in optical fibers [1] remained a phenomenon studied

by a small set of researchers for several years [2]. The demonstration of transverse

holographic writing [3] of gratings using two interfering beams external to the fiber

changed this and generated a lot of interest. The extension of this method to fabricate

reflection gratings around 1500nm using an ultraviolet laser [4] made it attractive for

telecommunication applications. Several advances [5] [6] [7] led to improvements in

reflectivity of fiber Bragg gratings paving the way for their widespread use in

telecommunication as wavelength division multiplexer, fiber lasers and dispersion

compensators [8] [9].

The applications of FBGs in sensors was identified quite early for strain and temperature

sensing [10] and for special applications like high pressure [11] and magnetic field [12]

sensing. Even though the initial applications [13] [14] did not gain adoption to the same

extent as in telecommunication, this has changed over the years with applications [15]

[16] [17] where the advantages of fiber Bragg grating sensors can be exploited to the

maximum [18] [19]. Further, the applicability of FBGs in large scale distributed sensing

has been exploited for structural health monitoring [20] while their small size,



36

multiplexing capabilities and electromagnetic immunity have led to several applications

in biomedical devices, biomechanics and medical robotics [21].

A fiber Bragg grating (FBG) consists of a periodic modulation of the refractive index in

the core of a single mode optical fiber [22]. Each grating plane scatters light guided

along the core of the fiber. When the Bragg condition is satisfied, reflected light from

each subsequent plane adds constructively to form a back reflected peak. The center

wavelength of the reflected peak is a function of the grating parameters [23].

Figure 2.1 Reflected peak and transmitted spectrum for a fiber Bragg grating

The center wavelength of the reflected peak is given by

λB = 2 neff Λ --- (2.1)

where neff is the effective refractive index of the fiber core and Λ is the grating period.

The simplest sensing principle of fiber Bragg gratings is based on the shift in the Bragg

wavelength due to variation in the periodic spacing between the grating planes and the

effective refractive index in response to an external perturbation such as strain and

temperature changes.



37

∆𝜆𝐵 = 2 (𝛬𝜕𝑛𝑒𝑓𝑓

𝜕𝑙+ 𝑛𝑒𝑓𝑓

𝜕𝛬

𝜕𝑙) ∆𝑙 + 2 (𝛬

𝜕𝑛𝑒𝑓𝑓

𝜕𝑇+ 𝑛𝑒𝑓𝑓

𝜕𝛬

𝜕𝑇) ∆𝑇 --- (2.2)

The first term corresponds to the strain response of the sensor and the second term to

the temperature response. Equation (2.2) can be reduced to

∆𝜆𝐵

𝜆𝐵= (1 − 𝑝𝑒)𝜀𝑧 + (𝛼𝑛 + 𝛼Λ)∆𝑇 --- (2.3)

Where 𝑝𝑒 is the effective photo-elastic constant for the optical fiber (≈0.22), 𝛼𝑛 is the

thermo-optic coefficient (≈ 8.6x10-6 for germania doped silica fiber), 𝛼Λ is the thermal

expansion coefficient of the fiber (≈ 0.55x10-6 for germania doped silica fiber) and 𝜀𝑧

is the longitudinal strain.

2.2 Sensor Fabrication

Interferometric methods of fabrication spurred active research in FBGs. However,

it is the phase mask technique of fabrication [24] [25] that has provided an

effective and robust method enabling large scale fabrication of high reflectivity

gratings in a reproducible way. All FBG sensors used in the experiments presented

in this thesis have been fabricated using the phase mask technique.

The phase mask is a periodic surface-relief pattern with period Λpm etched into

fused silica [26] and acts as a diffractive optical element. The profile is selected

such that the zeroth order diffracted beam is suppressed and the ±1 first orders

are maximized. The interference of the ±1 first orders produces a near field fringe

pattern whose period (Λ) is half that of the phase mask.



38

Figure 2.2 Schematic diagram of the phase mask method [9]

The details of the setup used for fabrication of sensors used for experiments in this

thesis is available in Appendix A.

2.3 Use in Biomedical Application

FBG sensors have been used for several biomedical applications where their small size,

multiplexing capability, electromagnetic immunity and high sensitivity provide a

significant advantage. The applications can be classified into the following areas based

on application areas:

i) MRI compatible devices

ii) Biomechanics

iii) Biosensors

iv) Smart end effectors for minimally invasive procedures

v) Guidance for minimally invasive procedures

The MRI compatibility of FBG sensors provide them a unique advantage which has

been exploited for monitoring respiration and cardiac activity in an MRI environment

[27] [28]. Different methods have been applied to increase the sensitivity and for signal



39

analysis based on sensor placement. These have even been extended to create wearable

sensors in the form of smart textiles [29] [30] which could make their use easier in an

MRI environment. FBG sensors have been integrated into needle holders used for

brachytherapy [31] for use in an MRI environment where standard force and torque

sensors cannot be used.

FBG sensors provide a good option for strain measurements in biomechanics and

rehabilitation. Their efficacy during studies involving measurement of tendon and

ligament strains and even detection of ligament deformation has been demonstrated

[32]. They have also been used for measurement of plantar strains and postural stability

[33]. In the area of prosthetics, they have been used for interface pressure measurement

between residual limb and prosthetic socket [34]. Additionally, they have found use in

monitoring decalcification of bones [35] and bone deformation [36]. One of the major

advantages of FBG sensors is the possibility to embed them during processing of

materials. This has been exploited for characterization of bone cement used for

cemented joint arthroplasty [37].

Etched FBG sensors have been used as biosensors by detecting the change in Bragg

wavelength due to refractive index change in the surrounding medium. They can act as

high sensitivity biosensors. This has been used to detect hybridization of DNA [38].

Etched FBG sensors coated with graphene oxide have been used for high sensitivity

detection of C-reactive protein [39] and glucose and glycated hemoglobin [40]. Other

recent applications of bare FBG sensors as biosensors include real time detection of

Escherichia coli [41] and as a biosensor with osteoblast cytocompatibility for possible

in-vivo applications in bone mechanical dynamics [42].



40

The application areas which have seen a significant amount of research activity

involving FBG sensors in recent years, apart from structural health monitoring, have

been related to the use of FBG sensors in smart instruments and end effectors with

sensing capabilities for use in medical procedures and in guidance and assistance for

minimally invasive procedures. It is evident that the small size, high multiplexing

capabilities and ability to withstand sterilization procedures provide them a significant

advantage for these applications. They have been integrated into forceps used in

minimally invasive surgery to determine grasping forces [43] and in scissor blades to

determine cutting forces [44]. Their effectiveness for force sensing has been

demonstrated for a complex procedure like vitro-retinal microsurgery [45] [46] [47].

They have been used to monitor temperature and contact force during radiofrequency

ablation [48]. Complex predictions like possibility of perforation during cardiac RF

ablation due to excessive force has also been attempted [49]. Higher acceptance of

minimally invasive procedures and robotic assistance in recent years have been

important factors for increase in research interest in the area. Robot assisted needle

steering [50] has motivated research towards shape estimation of flexible needles made

of nitinol [51] [52] .

2.4 Thesis contributions towards novel smart

instrument design using FBG sensors

Smart biomedical instruments incorporate sensing elements combined with a

computational system to provide improved usability and assistance. Chapter three of

this thesis focusses on the design of a smart needle which can determine the transition

through tissue layers. Most of the prior work in the area of smart needles has been

around solid flexible needles made using nitinol on which FBG sensors have been

bonded [51] [52] [53] [54]. These FBG sensors have been used to determine the needle



41

bending. This can be used for some of the minimally invasive surgical procedures.

However, a range of minimally invasive surgical and diagnostic procedures like

biopsies and radiological procedures use hollow needles. The work presented in this

thesis focusses on hollow needles for such procedures. The benefits of small size and

multiplexing capabilities of FBG sensors have been used to design a smart needle which

can determine the transition through tissue layers to enable better positioning of the

needle tip.

Shape estimation of flexible instruments including endoscopes and needles has been an

active research area in recent years. However, most of the work in this area in recent

years has focused either on using FBG sensors bonded on nitinol wires [51] [52] [53]

or on using FBG sensors fabricated in multi core fibers [55]. Chapter four of this thesis

demonstrates the limitations of using nitinol wires for this purpose. Using multi-core

fibers is promising. However, the sensor fabrication becomes more complex and special

3-D waveguides are required to couple the multi-core fibers to single mode fibers which

can be used with the FBG interrogators. An alternative method based on FBG sensors

embedded in a biocompatible polymer filled tube has been presented in this thesis. This

allows the use of standard phase mask technique for fabricating sensor arrays on single

mode fibers. A computational technique is presented for handling the non-linear

response of the sensors for varying bending radii.



42

References

[1] K. O. Hill, Y. Fujii, D. C. Johnson and B. S. Kawasaki, "Photosensitivity in optical

fiber waveguides: Application to reflection filter fabrication," Applied

physics letters, vol. 32, pp. 647-649, 1978.

[2] K. O. Hill, "Photosensitivity in optical fiber waveguides: From discovery to

commercialization," IEEE Journal of Selected Topics in Quantum

Electronics, vol. 6, pp. 1186-1189, 2000.

[3] G. Meltz, W. W. Morey and W. H. Glenn, "Formation of Bragg gratings in optical

fibers by a transverse holographic method," Optics letters, vol. 14, pp. 823-

825, 1989.

[4] R. Kashyap, J. R. Armitage, R. Wyatt, S. T. Davey and D. L. Williams, "All-fibre

narrowband reflection gratings at 1500 nm," Electronics Letters, vol. 26, pp.

730-732, 1990.

[5] P. J. Lemaire, R. M. Atkins, V. Mizrahi and W. A. Reed, "High pressure H/sub

2/loading as a technique for achieving ultrahigh UV photosensitivity and

thermal sensitivity in GeO/sub 2/doped optical fibres," Electronics Letters,

vol. 29, pp. 1191-1193, 1993.



43

[6] D. L. Williams, B. J. Ainslie, J. R. Armitage, R. Kashyap and R. Campbell,

"Enhanced UV photosensitivity in boron codoped germanosilicate fibres,"

Electronics Letters, vol. 29, pp. 45-47, 1993.

[7] J.-L. Archambault, L. Reekie and P. S. J. Russell, "100% reflectivity Bragg

reflectors produced in optical fibres by single excimer laser pulses,"

Electronics Letters, vol. 29, pp. 453-455, 1993.

[8] C. R. Giles, "Lightwave applications of fiber Bragg gratings," Journal of

Lightwave Technology, vol. 15, pp. 1391-1404, 1997.

[9] R. Kashyap, Fiber bragg gratings, Academic press, 1999.

[10] W. W. Morey, G. Meltz and W. H. Glenn, "Fiber optic Bragg grating sensors," in

OE/FIBERS'89, 1990.

[11] M. G. Xu, L. Reekie, Y. T. Chow and J. P. Dakin, "Optical in-fibre grating high

pressure sensor," Electronics letters, vol. 29, pp. 398-399, 1993.

[12] A. D. Kersey and M. J. Marrone, "Fiber Bragg grating high-magnetic-field probe,"

in 10th Optical Fibre Sensors Conference, 1994.

[13] Y.-J. Rao, "Fiber Bragg grating sensors: principles and applications," in Optical

fiber sensor technology, Springer, 1998, pp. 355-379.

[14] A. D. Kersey, M. A. Davis, H. J. Patrick, M. LeBlanc, K. P. Koo, C. G. Askins,

M. A. Putnam and E. J. Friebele, "Fiber grating sensors," Journal of

lightwave technology, vol. 15, pp. 1442-1463, 1997.



44

[15] B. Lee, "Review of the present status of optical fiber sensors," Optical fiber

technology, vol. 9, pp. 57-79, 2003.

[16] L. S. Grattan and B. T. Meggitt, Optical fiber sensor technology: advanced

applications-Bragg gratings and distributed sensors, Springer Science &

Business Media, 2013.

[17] A. D. Kersey, "30 Years of Fiber Bragg Grating Sensor Technology," in Bragg

Gratings, Photosensitivity, and Poling in Glass Waveguides, 2016.

[18] A. Mendez, "Fiber Bragg grating sensors: a market overview," in Third European

Workshop on Optical Fibre Sensors, 2007.

[19] A. Cusano, A. Cutolo and J. Albert, Fiber Bragg grating sensors: recent

advancements, industrial applications and market exploitation, Bentham

Science Publishers, 2011.

[20] M. Majumder, T. K. Gangopadhyay, A. K. Chakraborty, K. Dasgupta and D. K.

Bhattacharya, "Fibre Bragg gratings in structural health monitoring—Present

status and applications," Sensors and Actuators A: Physical, vol. 147, pp.

150-164, 2008.

[21] V. Mishra, N. Singh, U. Tiwari and P. Kapur, "Fiber grating sensors in medicine:

Current and emerging applications," Sensors and Actuators A: Physical, vol.

167, pp. 279-290, 2011.



45

[22] A. Othonos and K. Kalli, Fiber Bragg gratings: fundamentals and applications in

telecommunications and sensing, Artech House, 1999.

[23] K. O. Hill and G. Meltz, "Fiber Bragg grating technology fundamentals and

overview," Journal of lightwave technology, vol. 15, pp. 1263-1276, 1997.

[24] K. O. Hill, B. Malo, F. Bilodeau, D. C. Johnson and J. Albert, "Bragg gratings

fabricated in monomode photosensitive optical fiber by UV exposure

through a phase mask," Applied Physics Letters, vol. 62, pp. 1035-1037,

1993.

[25] D. Z. Anderson, V. Mizrahi, T. Erdogan and A. E. White, "Production of in-fibre

gratings using a diffractive optical element," Electronics Letters, vol. 29, pp.

566-568, 1993.

[26] Y. Qiu, Y. Sheng and C. Beaulieu, "Optimal phase mask for fiber Bragg grating

fabrication," Journal of lightwave technology, vol. 17, p. 2366, 1999.

[27] Ĺ. Lukasz Dziuda, F. W. Skibniewski, M. Krej and P. M. Baran, "Fiber Bragg

grating-based sensor for monitoring respiration and heart activity during

magnetic resonance imaging examinations," Journal of biomedical optics,

vol. 18, pp. 57006-57006, 2013.

[28] A. F. Silva, J. P. Carmo, P. M. Mendes and J. H. Correia, "Simultaneous cardiac

and respiratory frequency measurement based on a single fiber Bragg grating

sensor," Measurement Science and Technology, vol. 22, p. 075801, 2011.



46

[29] J. Witt, F. Narbonneau, M. Schukar, K. Krebber, J. De Jonckheere, M. Jeanne, D.

Kinet, B. Paquet, A. Depre, L. T. D'Angelo and others, "Medical textiles with

embedded fiber optic sensors for monitoring of respiratory movement," IEEE

sensors journal, vol. 12, pp. 246-254, 2012.

[30] J. P. Carmo, A. M. F. Silva, R. P. Rocha and J. H. Correia, "Application of fiber

Bragg gratings to wearable garments," IEEE Sensors Journal, vol. 12, pp.

261-266, 2012.

[31] R. Monfaredi, R. Seifabadi, G. Fichtinger and I. Iordachita, "Design of a

decoupled MRI-compatible force sensor using fiber Bragg grating sensors

for robot-assisted prostate interventions," in SPIE Medical Imaging, 2013.

[32] L. Ren, G. Song, M. Conditt, P. C. Noble and H. Li, "Fiber Bragg grating

displacement sensor for movement measurement of tendons and ligaments,"

Applied optics, vol. 46, pp. 6867-6871, 2007.

[33] A. S. G. Prasad, S. N. Omkar, H. N. Vikranth, V. Anil, K. Chethana and S.

Asokan, "Design and development of Fiber Bragg Grating sensing plate for

plantar strain measurement and postural stability analysis," Measurement,

vol. 47, pp. 789-793, 2014.

[34] E. A. Al-Fakih, N. A. A. Osman, F. R. M. Adikan, A. Eshraghi and P. Jahanshahi,

"Development and validation of fiber Bragg grating sensing pad for interface

pressure measurements within prosthetic sockets," IEEE Sensors Journal,

vol. 16, pp. 965-974, 2016.



47

[35] V. Mishra, N. Singh, D. V. Rai, U. Tiwari, G. C. Poddar, S. C. Jain, S. K. Mondal

and P. Kapur, "Fiber Bragg grating sensor for monitoring bone

decalcification," Orthopaedics & Traumatology: Surgery & Research, vol.

96, pp. 646-651, 2010.

[36] T. Fresvig, P. Ludvigsen, H. Steen and O. Reikerås, "Fibre optic Bragg grating

sensors: an alternative method to strain gauges for measuring deformation in

bone," Medical engineering & physics, vol. 30, pp. 104-108, 2008.

[37] C. Frias, O. Frazão, S. Tavares, A. Vieira, A. T. Marques and J. Simões,

"Mechanical characterization of bone cement using fiber Bragg grating

sensors," Materials & Design, vol. 30, pp. 1841-1844, 2009.

[38] A. N. Chryssis, S. S. Saini, S. M. Lee, H. Yi, W. E. Bentley and M. Dagenais,

"Detecting hybridization of DNA by highly sensitive evanescent field etched

core fiber Bragg grating sensors," IEEE Journal of selected topics in

Quantum Electronics, vol. 11, pp. 864-872, 2005.

[39] S. Sridevi, K. S. Vasu, S. Asokan and A. K. Sood, "Sensitive detection of C-

reactive protein using optical fiber Bragg gratings," Biosensors and

Bioelectronics, vol. 65, pp. 251-256, 2015.

[40] S. Sridevi, K. S. Vasu, S. Sampath, S. Asokan and A. K. Sood, "Optical detection

of glucose and glycated hemoglobin using etched fiber Bragg gratings coated

with functionalized reduced graphene oxide," Journal of biophotonics, 2015.



48

[41] R. Srinivasan, S. Umesh, S. Murali, S. Asokan and S. Siva Gorthi, "Bare fiber

Bragg grating immunosensor for real-time detection of Escherichia coli

bacteria," Journal of biophotonics, 2016.

[42] L. Carvalho, N. J. Alberto, P. S. Gomes, R. N. Nogueira, J. L. Pinto and M. H.

Fernandes, "In the trail of a new bio-sensor for measuring strain in bone:

Osteoblastic biocompatibility," Biosensors and Bioelectronics, vol. 26, pp.

4046-4052, 2011.

[43] S.-C. Lim, H.-K. Lee and J. Park, "Grip force measurement of forceps with fibre

Bragg grating sensors," Electronics Letters, vol. 50, pp. 733-735, 2014.

[44] D. Callaghan, "Force Sensing Surgical Scissor Blades using Fibre Bragg Grating

Sensors," 2013.

[45] I. Iordachita, Z. Sun, M. Balicki, J. U. Kang, S. J. Phee, J. Handa, P. Gehlbach and

R. Taylor, "A sub-millimetric, 0.25 mN resolution fully integrated fiber-optic

force-sensing tool for retinal microsurgery," International journal of

computer assisted radiology and surgery, vol. 4, pp. 383-390, 2009.

[46] X. He, M. A. Balicki, J. U. Kang, P. L. Gehlbach, J. T. Handa, R. H. Taylor and

I. I. Iordachita, "Force sensing micro-forceps with integrated fiber bragg

grating for vitreoretinal surgery," in SPIE BiOS, 2012.

[47] B. Gonenc, M. A. Balicki, J. Handa, P. Gehlbach, C. N. Riviere, R. H. Taylor and

I. Iordachita, "Preliminary evaluation of a micro-force sensing handheld



49

robot for vitreoretinal surgery," in Intelligent Robots and Systems (IROS),

2012 IEEE/RSJ International Conference on, 2012.

[48] D. Tosi, E. G. Macchi, G. Braschi, M. Gallati, A. Cigada, S. Poeggel, G. Leen and

E. Lewis, "Monitoring of radiofrequency thermal ablation in liver tissue

through fibre Bragg grating sensors array," Electronics Letters, vol. 50, pp.

981-983, 2014.

[49] S. C. M. Ho, M. Razavi, A. Nazeri and G. Song, "FBG sensor for contact level

monitoring and prediction of perforation in cardiac ablation," Sensors, vol.

12, pp. 1002-1013, 2012.

[50] K. B. Reed, A. Majewicz, V. Kallem, R. Alterovitz, K. Goldberg, N. J. Cowan

and A. M. Okamura, "Robot-assisted needle steering," IEEE robotics &

automation magazine, vol. 18, pp. 35-46, 2011.

[51] Y.-L. Park, S. Elayaperumal, B. Daniel, S. C. Ryu, M. Shin, J. Savall, R. J. Black,

B. Moslehi and M. R. Cutkosky, "Real-time estimation of 3-D needle shape

and deflection for MRI-guided interventions," IEEE/ASME Transactions On

Mechatronics, vol. 15, pp. 906-915, 2010.

[52] M. Abayazid, M. Kemp and S. Misra, "3d flexible needle steering in soft-tissue

phantoms using fiber bragg grating sensors," in Robotics and Automation




50

[53] R. J. Roesthuis, M. Kemp, J. J. Dobbelsteen and S. Misra, "Three-dimensional

needle shape reconstruction using an array of fiber bragg grating sensors,"

IEEE/ASME Transactions on mechatronics, vol. 19, pp. 1115-1126, 2014.

[54] K. R. Henken, J. Dankelman, J. J. Dobbelsteen, L. K. Cheng and M. S. Heiden,

"Error analysis of FBG-based shape sensors for medical needle tracking,"

IEEE/ASME Transactions on mechatronics, vol. 19, pp. 1523-1531, 2014.

[55] J. P. Moore and M. D. Rogge, "Shape sensing using multi-core fiber optic cable

and parametric curve solutions," Optics Express, vol. 20, pp. 2967-2973,

2012.

Chapter 3 Estimating Needle Transitions through

Tissue Layers

51

Chapter 3

Estimating Needle Transitions through

Tissue Layers

3.1 Introduction

Needles are used in a variety of medical procedures ranging from simple subcutaneous

and intravenous administration of medicines to complex neurological procedures.

Minimally invasive diagnostic and therapeutic procedures such as biopsies and

radiological seed implantation (brachytherapy) which have had tremendous impact on

cancer management are highly dependent on the use of needles.

Needles are available in a wide variety of geometries. Some of the most common

commercially available needles have variants of bevel shaped cutting tips

(brachytherapy and spinal needles) and symmetrical multi-plane cutting tips (Franseen

needles used in lung biopsy). The tip geometries can be seen in figure 3.1.

Needles also have a wide range of outer diameter. The needle sizes are usually specified

in terms of gauges. The gauge size (G) is inversely related to the needle diameter. For

example, a 26 gauge needle has a smaller diameter than an 18 gauge needle. Table 3.1

provides the nominal inner and outer diameters for common needle gauges.


Tissue Layers

52

(a)

(b)

Figure 3.1 (a) Bevel tipped and (b) Franseen needle

Table 3.1 Inner and outer diameters for common needle gauge sizes

Needle Gauge Nominal Outer Diameter

(mm)

Nominal Inner Diameter

(mm)

14 2.11 1.6

16 1.65 1.19

18 1.27 0.84

20 0.91 0.6

21 0.82 0.51

22 0.72 0.41

24 0.57 0.31

26 0.46 0.26

28 0.36 0.18

30 0.31 0.16


Tissue Layers

53

18G needles are commonly used for brachytherapy and epidural procedures. All

experiments reported in this work have been done using 18G bevel tipped needles

(Quincke) [1].

The advancements in use of robotic and robot-assisted procedures have led to significant

interest in evaluation and modelling of needle tissue interactions [2] [3] [4] [5]. Needle

insertion into tissues is characterized by multiple stages.

(a)

(b)

(c)

Figure 3.2 Stages of needle insertion: (a) tissue deflection (b) rupture (c) movement

inside tissue

ftip

ftip

ftip ffriction

ffriction


Tissue Layers

54

In the first stage as shown in figure 3.2 (a), the tissue undergoes deflection. This is

followed by tissue rupture and the tip enters the tissue as shown in figure 3.2 (b). After

this, the needle tip and the shaft move inside the tissue as shown in figure 3.2 (c).

When the needle moves inside the tissue, it experiences force at the tip as well as

frictional force. There have been several studies to characterize the forces during

deflection as well as the frictional forces. The initial work in the area [6] [7] used

homogeneous linear models for tissues. These have been used to create virtual

simulations for needle insertion. Even though they enable creation of visualizations,

these models have limitations as tissues are inhomogeneous and exhibit nonlinear

viscoelastic behavior. In later work, Okamura et.al. [8] have provided a model using

multiple stages and split the forces into force due to stiffness, frictional forces and

cutting forces. A polynomial model for deflection has been used. However, experiments

with bovine liver has shown high variation in the estimated parameters. This could be

the result of high variations in tissue properties and the presence of vasculature.

In recent years, there has been higher focus on robotic insertion of needles [9] [10] [11]

[12]. These have used measurement of forces at the base of needle holder. Several

experiments have used specially created nitinol (a nickel-titanium alloy) needles which

have high flexibility [13]. The focus has been mostly on needle steering. Several

experiments have used FBG sensors for estimating the shape of these needles [14] [15]

[16] [17] [18]. However, work on hollow needles has been limited [19]. Hollow needles

have different cutting forces due to the tip geometry compared to solid needles.

Estimating forces directly at the needle tip is interesting in this case. This can be used

to avoid modelling the frictional forces which can have high variations. It also allows

their usage in procedures where direct estimation of forces at the base of a large needle

holder is not possible as in the case of NOTES [20] [21] [22].


Tissue Layers

55

3.2 Needle and Sensor Characteristics

The most important parameters to consider for hollow needles are the needle diameter

and needle tip geometry. As indicated before, all experiments presented here have been

carried out using 18G bevel tipped needles.

Before deciding on placement of sensors, FEM simulation has been performed using an

approximate model in COMSOL and applying a fixed load at the cutting edge of the

needle as shown in figures 3.3 and 3.4. This is done to check if the amount of strain that

would be encountered along the axis of the needle can be measured using the sensing

setup.

Figure 3.3 Simulation configuration with one fixed end

In the simulation results, as seen in figure 4, for a load of 0.1N applied axially at the tip,

a strain of approximately 5 micro-strains has been observed at regions 1-2mm away

from the tip along the axis.


Tissue Layers

56

Figure 3.0.4 Load of 0.1N applied at cutting edge

For constant temperature the strain response of FBG sensors can be reduced to

∆ 𝜆𝑏

𝜆𝑏 = (1-𝑝𝑒)𝜀𝑧 = (1-0.213) 𝜀𝑧 = 0.787𝜀𝑧 --- (3.1)

where pe is the photo-elastic constant and 𝜀𝑧 is the longitudinal strain.

The strain of 5 micro-strains for a 0.1N load as seen in the simulation results corresponds

to a wavelength shift of 3.94 pico-meter. This can be detected by the FBG interrogation

systems used for the experiments presented here.

In the present work, FBG sensors have been fabricated in a 125 micron germanosilicate

optical fiber (Nufern) using phase mask method. Phase mask method provides high

repeatability and control on the Bragg wavelength. However, one challenge with the

method is replacement of phase masks for fabricating sensor arrays. Two sensors close

to each other have been fabricated using pre-stretching of the fiber for the second sensor

and translating the fiber as shown in figure 3.5. This allows easy fabrication of sensors

with a Bragg wavelength separation in the range of 1.5-2 nm.


Tissue Layers

57

(a)

(b)

Figure 3.5 Fabrication of two FBG sensors close to each other by translating and pre-

stretching the fiber

The gauge length of each sensor is 3mm and the separation between the two sensors is

about 1.5 cm. The fiber has been bonded inside the needle lumen along the axis as shown

in figure 3.6. The polymer jacket on the fiber has been completely removed in the region

around and between the sensors. This is done to avoid any uneven bonding. During

bonding, the fiber has been maintained in a very slight tension to avoid any bending.

The tip of the fiber beyond the first senor has been cut after the completion of bonding.

UV Beam (248nm)

Translation Stage Fiber Core

Fiber Holder

First Sensor

UV Beam (248nm) Fiber Stretching


Tissue Layers

58

Figure 3.6 Needle with FBG sensors bonded

3.3 Needle Insertion Setup

A test setup has been designed for the controlled needle insertion and to compare forces

measured by the two FBG sensors in correspondence with the forces at the needle holder

base during the needle insertion into tissue-phantoms. It consists of a linear actuation

assembly, a force sensor, a displacement sensor and needle and tissue-phantom holders

as shown schematically in figure 3.7. The linear actuation is controlled manually or

programmed for specific insertion distances at varying speeds through a programmable

logic controller. An S-type strain gauge fixed to the needle holder assembly measures

the forces with a resolution of 1mN with a range of 0-10N. A syringe with the

instrumented needle is rigidly fixed to the holder. A Linear Variable Displacement

Transformer (LVDT) is used as an absolute position sensor. The tissue mimicking

phantoms are contained in a plastic tube rigidly held in the specimen holder. This is

required to prevent the phantom from moving along with the needle during insertion.

Optical Fiber

FBG Sensors To Interrogator

3mm

15mm


Tissue Layers

59

Fig 3.7 Schematic diagram of the needle insertion setup

(a) (b)

Figure 3.8 (a) Complete setup including needle translation stage, controller and data

acquisition setup and (b) Needle being inserted in the phantom with a camera placed

to monitor its progress inside the phantom


Tissue Layers

60

3.4 Polydimethylsiloxane (PDMS) Phantom

Polydimethylsiloxane (PDMS) is a silicone based polymer. It is optically clear which

has led to its usage as tissue mimicking phantom in optical coherence tomography [23]

[24]. Another major advantage is the ability to vary its mechanical properties in a

controlled manner by varying the cross-linking. Hence, it has also been used in studies

related to impact response of biological tissue [25]. The phantom is stable over longer

duration and can be stored easily at room temperature. It can be easily used at

temperatures around 37-40°C which is the usual range of body temperature without

problems related to loss of water and major changes in mechanical properties as

observed with gelatin based phantoms at such temperatures.

Sylgard 184 silicone elastomer kit has been used for preparing the phantoms. It consists

of a base elastomer and a curing agent. For all experiments, phantoms have been

prepared using base polymer to curing agent ratios of 10:0.6 and 10:0.3 and curing at

80°C.

3.5 Measurements and Results

3.5.2 Sensitivity Determination

In order to determine the sensitivity of the instrumented needle to external forces,

calibration has been performed by applying small external loads at the needle tip and

simultaneously measuring the strains measured by the two FBG sensors. The response

obtained is linear as seen in figure 3.9 and the sensitivity of the two sensors is obtained

from the slope of a linear least square fit to the data points.


Tissue Layers

61

Figure 3.9 Strain response of the two FBG sensors

Both loading and unloading experiments have been performed and no hysteresis effects

have been observed. The experiments have been repeated in the limit of 0-1.4N external

forces at the tip and no nonlinear effects are seen. The sensitivity of FBG sensor 1 (closer

to the tip) obtained is 67.3 micro-strains / N and that of FBG sensor 2 is 83.6 micro-

strains / N. The reason for difference in sensitivity is the slight bending at the tip at

higher loads, which acts against compression.

3.5.2 Transition through layers with different

stiffness

Since tumors can have a higher stiffness than healthy tissue [26], one of the goals in

needle penetration estimation is to determine transition from a layer of lower stiffness

to a layer of higher stiffness followed by transition to a layer of lower stiffness. This

includes transition estimation when the needle tip is already surrounded by tissue and

hence the frictional forces are acting on the needle. To analyze the effectiveness of using


Tissue Layers

62

force measurements at the tip for this situation, phantoms with five layers have been

created using PDMS by varying the ratio of base polymer to curing agent as shown in

figure 3.10.

X5 (mm) X4 (mm) X3 (mm) X2 (mm) X1 (mm)

Phantom1 4 7.5 5 8 4

Phantom2 3 8 4 7 4

Figure 3.10 Phantoms with five layers with alternating higher and lower stiffness

The higher stiffness layer is created using a base polymer to curing agent ratio of 10:0.6

and the lower stiffness layer is created using a ratio of 10:0.3. The phantom is created

by adding layers one at a time. The second layer is added after partial curing of first

layer and so on. This method enables creation of multi layered PDMS phantoms without

the need for plasma treatment of layer surfaces. The phantom is prepared inside a plastic

tube with sealing at one end to avoid motion of the phantom during needle insertion.

The needle insertion is made at 1mm/s, 2mm/s, 3mm/s and 4mm/s from the layer with

higher stiffness towards the next layers as shown in figure 3.11 and the strain responses

of the two sensors are recorded along with the needle displacement and force at the

needle holder base. The response of the two FBG sensors during transition through

different layers is shown in figure 3.12.

X1

10:0.3 10:0.6 10:0.3 10:0.6

Needle Insertion

Direction

X2 X3 X4 X5

10:0.6


Tissue Layers

63

(a)

(b)

(c)

(d)

(e)

(f)

Figure 3.11 Stages during needle insertion in the phantom

10:0.3

4mm 1mm 3mm

10:0.6 10:0.3 10:0.6

3mm

10:0.3

4mm 1mm 3mm

10:0.6 10:0.3 10:0.6

3mm

10:0.3

4mm 1mm 3mm

10:0.6 10:0.3 10:0.6

3mm

10:0.3

4mm 1mm 3mm

10:0.6 10:0.3 10:0.6

3mm

10:0.3

4mm 1mm 3mm

10:0.6 10:0.3 10:0.6

3mm

10:0.3

1mm 3mm 4mm

10:0.6 10:0.3 10:0.6

3mm


Tissue Layers

64

An important point to consider is the deformation of the phantoms during insertion. The

presence of alternative layers with higher and lower stiffness also leads to varying

amounts of deformation at different stages of insertion. This is reflected in the graph in

figure 3.12. Hence the points at which transitions occur do not match exact boundaries

as shown in figure 3.11. The deformation of first layer is significantly higher than that

of the third layer. The deformations also vary slightly with insertion velocity.

.

Figure 3.12 Response of the sensors during needle insertion

In order to analyze and use the response, it is required to find the features in the signal

which can be reliably used to determine transitions between layers in spite of variation

in needle insertion speed and depth of insertion. Strain gradients are one such feature.

Figure 3.13 shows the gradient of the strains and the force at the needle holder base

computed using central difference method. The gradients show that sensor one (closer

to the needle tip) responds faster during the transition between layers. Another


Tissue Layers

65

advantage of using the gradient of strains measured by sensor one is the zero crossing

of the gradients. These are the points where the needle tip moves inside the layers with

higher stiffness (points ‘b’ and ‘e’) after rupture leading to a reduction in strains for a

small insertion distance. The effect of deformation is evident by comparing the initial

boundaries of the layers in figure 3.11 and the points ‘b’ and ‘e’ in figure 3.13. Using

the gradient allows the identification of transitions without depending directly on the

knowledge of the thickness of the layers or on the estimation of the deformation and the

subsequent relaxation.

Figure 3.13 Gradients of the strains and the force at the needle holder base

The forces are known to vary even for the same tissues for varying insertion rates.

Hence, it is important to show that the use of gradients is valid for varying needle

insertion rates. The strains and the gradients for insertion at 3mm/s can be seen

respectively in figures 3.14 and 3.15. It can be seen that the gradients of strains measured

by sensor 1 still show a similar pattern and can be used to identify the transitions.


Tissue Layers

66

Figure 3.14 Strains for insertion at 3mm/s

Figure 3.15 Gradients for insertion at 3mm/s

The response of the sensors for all four needle insertion rates and for both phantoms can

be seen in figure 3.16 and 3.17 respectively. The figures show that the nature of response

is similar across needle insertion rates and phantoms with varying thickness of the

layers. The small variations in first transition point based on insertion velocity can also

be seen.


Tissue Layers

67


needle holder base for insertion in phantom 1 at varying needle insertion rates


Tissue Layers

68


needle holder base for insertion in phantom 2 at varying needle insertion rates


Tissue Layers

69

The measurements show that absolute values of the strains and forces vary across

insertion velocities and phantoms. However, the trends in variation in strains remain the

same. These trends are reflected in the gradients which in turn follow similar patterns

and can be used for transition estimation.

3.6 Temperature Compensation

For experiments using phantoms, there is no temperature change during needle

insertion. However, during insertion in real tissue, there can be slight variations in the

temperature. Hence compensation is required to handle FBG sensor’s cross sensitivity

to temperature. This is achieved by using the configuration as shown in figure 3.18.

Figure 3.18 Additional hollow needle with FBG sensor used instead of stylet

A needle stylet is a solid rod inserted in the needle lumen to prevent any occlusion

during insertion. In the modified design used for temperature compensation, the solid

stylet has been replaced by a hollow needle with sealed tip. Another fiber with an FBG

sensor is bonded close to the tip inside this needle. The outer diameter of this needle is

chosen close to the inner diameter of the actual needle. During insertions, no strain is

induced on this FBG sensor. However, since it is located at the same place as the FBG


Tissue Layers

70

sensor 1, it responds to any temperature changes. This enables the elimination of cross

sensitivity to temperature.

The temperature sensitivity of a 1550nm FBG sensors in a germanium doped silica core

fiber is approximately 14 pm/°C. The temperature response of the two sensors has been

analyzed by placing the needle on a controlled heating plate and applying a step change

from 36°C to 38°C. The temperature of the hot plate rises gradually and then settles

around the final value. The temperature value is chosen close the usual temperature of

human body. The response of the two sensors is as seen in figure 3.19. The slightly

higher settling point for a 2°C change is because of the fact that the final temperature of

the heating plate slightly overshoots 38°C due to inaccuracies in the control method.

Figure 3.19 Temperature response of actual FBG sensor and the compensation FBG

sensor


Tissue Layers

71

During the period where temperature is rising, the maximum error between the actual

sensor on the 18 gauge needle and the compensation sensor in the inside needle is 6pm

and the average error is 3.5pm. Since the wavelength shifts due to strain during needle

insertion are much higher and strain variations are much faster, the compensation can

be used to account for changes in temperature.

3.7 Insertion in heated chicken tissue

To validate that the temperature compensation works in a real tissue, insertions have

been made in a chicken tissue which has been heated using the setup as shown in figure

3.20. The purpose of heating the chicken tissue is to create a temperature gradient.

Figure 3.20 Setup for insertions in heated chicken tissue

The response of senor bonded on the 18-gauge needle is shown in figure 3.21. The effect

of temperature can be clearly seen as the wavelength shift is positive and it continues to

rise because of the temperature till it reaches the settling point. If temperature effect is


Tissue Layers

72

not compensated, it would appear as a response to an axial force causing elongation

instead of compression of the needle during insertion.

Figure 3.21 Temperature induced wavelength shift during needle insertion

The response of the FBG sensor bonded on the measurement needle (18G) and of the

FBG sensor on the inside needle which acts as the stylet can be seen in figure 3.22 (a).

It can be observed that the sensor on the inside needle responds only to the temperature

change. This allows to compensate for temperature induced effects on the measurements

by 18G needle which measures both temperature and strain induced effects during

insertion. This can be done by subtracting the response of the sensor on the inside needle

from that of the sensor on the 18G needle. The compensated response is as shown in

figure 3.22(b).


Tissue Layers

73

(a)

Figure 3.22 (a) Response of the sensor bonded to the 18G needle and that of the

inside needle during insertion in heated tissue and (b) the temperature compensated

response of the sensor bonded to the 18G needle

The gradients of the strain response after applying temperature compensation has been

plotted in figure 3.23 to compare it with those obtained using the insertions in the PDMS

phantoms. The point before needle puncture ‘(a)’ and the point when the needle

punctures the first layer and goes inside tissue ‘(b)’ can be identified in the gradients.


Tissue Layers

74

The movement inside the tissue ‘(c)’ and transition inside the second layer ‘(d)’ can also

be seen.

Figure 3.23 Gradients of strain measured by senor 1 bonded on the 18G needle after

temperature compensation

3.8 Summary

The results from the experiments show that it is possible to detect the different stages

of needle penetration using Fiber Bragg Grating sensors placed close to the needle tip.

The advantages of the small sensor size and high sensitivity are evident. The use of

gradients of the strain measured by sensor 1 in the experiments provides a means to

determine the transitions without use of exact thresholds and slopes. It also overcomes

the challenges posed by the variation in absolute values and the rates of changes of

strains for insertions at varying rates. This would allow its usage in flexible instruments

without the need of force measurements at the needle holder base. This would also

enable developing assistive systems for procedures which involve precise positioning

of needles beyond certain organ or tissue boundaries.


Tissue Layers

75

References

[1] N. Calthorpe, "The history of spinal needles: getting to the point," Anaesthesia,

vol. 59, pp. 1231-1241, 2004.

[2] R. Alterovitz, K. Goldberg, J. Pouliot, R. Taschereau and I.-C. Hsu, "Needle

insertion and radioactive seed implantation in human tissues: Simulation and

sensitivity analysis," in Robotics and Automation, 2003. Proceedings.

ICRA'03. IEEE International Conference on, 2003.

[3] M. Mahvash and P. E. Dupont, "Mechanics of dynamic needle insertion into a

biological material," IEEE Transactions on Biomedical Engineering, vol. 57,

pp. 934-943, 2010.

[4] K. B. Reed, A. Majewicz, V. Kallem, R. Alterovitz, K. Goldberg, N. J. Cowan

and A. M. Okamura, "Robot-assisted needle steering," IEEE robotics \&

automation magazine, vol. 18, pp. 35-46, 2011.

[5] G. Wan, Z. Wei, L. Gardi, D. B. Downey and A. Fenster, "Brachytherapy needle

deflection evaluation and correction," Medical physics, vol. 32, pp. 902-909,

2005.

[6] S. P. DiMaio and S. E. Salcudean, "Needle insertion modeling and simulation,"

IEEE Transactions on robotics and automation, vol. 19, pp. 864-875, 2003.


Tissue Layers

76

[7] S. P. DiMaio and S. E. Salcudean, "Interactive simulation of needle insertion

models," IEEE transactions on biomedical engineering, vol. 52, pp. 1167-

1179, 2005.

[8] A. M. Okamura, C. Simone and M. D. Oĺeary, "Force modeling for needle

insertion into soft tissue," IEEE transactions on biomedical engineering, vol.

51, pp. 1707-1716, 2004.

[9] A. Majewicz, T. R. Wedlick, K. B. Reed and A. M. Okamura, "Evaluation of

robotic needle steering in ex vivo tissue," in Robotics and Automation


[10] S. Badaan, D. Petrisor, C. Kim, P. Mozer, D. Mazilu, L. Gruionu, A. Patriciu, K.

Cleary and D. Stoianovici, "Does needle rotation improve lesion targeting?,"

The International Journal of Medical Robotics and Computer Assisted

Surgery, vol. 7, pp. 138-147, 2011.

[11] S. E. Salcudean, T. D. Prananta, W. J. Morris and I. Spadinger, "A robotic needle

guide for prostate brachytherapy," in Robotics and Automation, 2008. ICRA

2008. IEEE International Conference on, 2008.

[12] A. Jahya, F. van der Heijden and S. Misra, "Observations of three-dimensional

needle deflection during insertion into soft tissue," in Biomedical Robotics

and Biomechatronics (BioRob), 2012 4th IEEE RAS & EMBS International

Conference on, 2012.


Tissue Layers

77

[13] J. van Gerwen and Dennis, J. Dankelman and J. van den Dobbelsteen and John,

"Needle--tissue interaction forces--A survey of experimental data," Medical

engineering \& physics, vol. 34, pp. 665-680, 2012.

[14] M. Abayazid, M. Kemp and S. Misra, "3d flexible needle steering in soft-tissue

phantoms using fiber bragg grating sensors," in Robotics and Automation






[16] R. J. Roesthuis, M. Kemp, J. van den Dobbelsteen and John and S. Misra, "Three-

dimensional needle shape reconstruction using an array of fiber bragg grating

sensors," IEEE/ASME Transactions on mechatronics, vol. 19, pp. 1115-

1126, 2014.

[17] H. Sadjadi, K. Hashtrudi-Zaad and G. Fichtinger, "Fusion of electromagnetic

trackers to improve needle deflection estimation: simulation study," IEEE


[18] K. R. Henken, J. Dankelman, J. van den Dobbelsteen and John, L. K. Cheng and

S. van der Heiden and Maurits, "Error analysis of FBG-based shape sensors

for medical needle tracking," IEEE/ASME Transactions on mechatronics,

vol. 19, pp. 1523-1531, 2014.


Tissue Layers

78

[19] J. Z. Moore, "Tissue Cutting in Needle Biopsy," 2011.

[20] I. Halim and A. Tavakkolizadeh, NOTES: The next surgical revolution?, Elsevier,

2008.

[21] S. Gillen, J. Kleeff, M. Kranzfelder, S. V. Shrikhande, H. Friess and H. Feussner,

"Natural orifice transluminal endoscopic surgery in pancreatic diseases,"

World Journal of Gastroenterology: WJG, vol. 16, p. 3859, 2010.

[22] C. F. Granberg, M. T. Gettman and others, "Instrumentation for natural orifice

translumenal endoscopic surgery and laparoendoscopic single-site surgery,"

Indian Journal of Urology, vol. 26, p. 385, 2010.

[23] F. Ayers, A. Grant, D. Kuo, D. J. Cuccia and A. J. Durkin, "Fabrication and

characterization of silicone-based tissue phantoms with tunable optical

properties in the visible and near infrared domain," in Biomedical Optics

(BiOS) 2008, 2008.

[24] G. Lamouche, B. F. Kennedy, K. M. Kennedy, C.-E. Bisaillon, A. Curatolo, G.

Campbell, V. Pazos and D. D. Sampson, "Review of tissue simulating

phantoms with controllable optical, mechanical and structural properties for

use in optical coherence tomography," Biomedical optics express, vol. 3, pp.

1381-1398, 2012.

[25] Z. I. Kalcioglu, R. A. Mrozek, R. Mahmoodian, M. R. VanLandingham, J. L.

Lenhart and V. a. K. J. Vliet, "Tunable mechanical behavior of synthetic


Tissue Layers

79

organogels as biofidelic tissue simulants," Journal of biomechanics, vol. 46,

pp. 1583-1591, 2013.

[26] V. Jalkanen, B. M. Andersson, A. Bergh, B. Ljungberg and O. A. Lindahl,

"Prostate tissue stiffness as measured with a resonance sensor system: a study

on silicone and human prostate tissue in vitro," Medical and Biological

Engineering and Computing, vol. 44, p. 593, 2006.


Tissue Layers

80

Chapter 4 Shape Estimation of Flexible Medical

Instruments

81

Chapter 4

Shape Estimation of Flexible Medical

Instruments

4.1 Introduction

Introduction of flexible medical instruments has provided a new paradigm for diagnosis

and treatment of several medical conditions. Flexible endoscopes are one of the most

prominently used flexible instruments. In general, flexible endoscopes consist of an

insertion tube, a connector section and a control section as shown in figure 4.1 (a). The

doctor typically guides the insertion tube using one hand while controlling knobs using

the other hand as shown in fig 4.1 (b).

(a)

(b)

Figure 4.1 (a) Schematic of Endoscope and (b) holding an endoscope for procedure


Instruments

82

The insertion tube is a flexible tube with working channels that enable passage of

accessories. The tip of the flexible tube is controlled using pull wires attached to it below

the protective sheath. The tube also has a working channel through which accessories

like graspers and needles can be passed and can be seen in the field of view of the

endoscope. The tip can be fixed in a position to perform an investigation on to use the

accessories. The connector provides connection to the electrical system, the light source,

the air and water system and the suction system. The control section has knobs to control

the bending at the distal end. Several types of scopes with varying tube lengths,

diameters and channel sizes are available [1] depending on the type of medical

procedures [2] for which they are used and the supplier. Table 4.1 provides a brief

categorization.

Table 4.1 Flexible endoscope types with typical dimensions

Endoscope Type Procedure

Gastroscope

Viewing: Forward Viewing

Diameter: 4.9-12.8mm

Channel size: 2.0-3.8mm

Insertion tube length: 925-

1100mm

Upper gastrointestinal endoscopy for

esophagus, stomach and duodenum for

detection of Barret’s esophagus, removal

of polyps, biopsy of gastrointestinal

cancers etc.

Colonoscope


Diameter: 11.1 - 15mm



1700mm

Colonoscopy for detection and treatment

of polyps, fissures, colorectal cancer etc.


Instruments

83

Sigmoidoscope


Diameter: 11.3 – 12.8mm



790mm

Sigmoidoscopy (for examining the lining

of the rectum and the large intestine) or

detection of polyps, rectal bleeding and

lesions

Duodenoscope





790mm

Endoscopic retrograde cholangio-

pancreatography (ERCP) to diagnose and

treat problems of the liver, pancreas and

gallbladder

Enteroscope





2200mm

Enteroscopy (for small intestine) to find

the source of internal bleeding

4.2 Need for shape estimation and assistance

In endoscopic procedures, the surgeon operates with a constrained view of the site as

seen through the tip of the scope. To reach a particular site during colonoscopy or upper

gastrointestinal endoscopy, the scope has to be navigated through a series of complex

bends. This leads to disorientation or ‘getting lost’ effect impeding the performance of

the endoscopist [3]. For example, in colonoscopy, the formation of loops is considered


Instruments

84

as one of the main difficulties related to manipulation of the scope for the endoscopist.

An example is shown in figure 4.2

Figure 4.2 Sketch of endoscope looping during colonoscopy

Typically landmarks are used by the endoscopist to navigate. However, they are not

always easy to identify and there can also be variation from patient to patient. The

endoscopist has to create a mental picture of the orientation of the scope to proceed with

the insertion. This is sometimes difficult [4] even for experienced endoscopists. Several

types of loops can be formed during endoscopy. One of the common ones is an ‘n –

loop’ as shown in fig 4.3 (a). The tip of the endoscope pushes against the colon lining

when pushed further in this situation and does not move forward. Pushing further exerts

force of the wall of the colon as shown in figure 4.3(b). The endoscopist can avoid the

formation of the loop by knowing the trajectory of the scope and the shape it takes. This

can be easily done by straightening the scope when the loop gets formed by pulling it

back and then reinserting it. This can be enabled by estimating the shape of the scope

and assisting the endoscopist through visual feedback.


Instruments

85

(a)

(b)

Figure 4.3 (a) Sketch showing formation of an ‘n’ loop and (b) pushing the scope further

pushes it against the wall of the colon instead of moving it forward

Apart from shape estimations and assistance during in-vivo procedures, training systems

for endoscopy provide another important area where shape estimation can be very

useful. Training on actual patients for endoscopy is a major constraint in practicing the

techniques. Even though ex-vivo and in-vivo training with pig models is possible [5],

this requires special handling at training hospitals. Several training systems for flexible

endoscopy have been developed over the years [6] [7] [8] [9]. These can be used with

real endoscopes and provide haptic feedback as shown in the schematic in figure 4.4.

These provide simulated visualizations as the endoscope is moved forward. However,

rigid systems with only forward insertion do not provide a realistic training

environment. Enabling the detection of bending of endoscopes can allow better

assistance during training through suggestions for pulling back the scope and

reinsertion. This can also help in simulating the effects of the scope pressing against the

colon walls causing stretching.


Instruments

86

Figure 4.4 Simple image showing an endoscopic training system (Image created as

part of ‘Cyber surgery and remote patient care’ project at Robert Bosch Center for

Cyber Physical Systems, Indian Institute of Science which has led to the startup

Mimyk [10])

4.3 Shape Estimation Methods

Direct viewing of the shape of the endoscope is possible using fluoroscopic methods

[11]. However, exposure to radiation leads to constraints in usage. Using it only for the

purpose of shape estimation is difficult to justify unless other benefits are available for

a complex surgical process. Also, it cannot be used continuously creating constraints

for its use in an assistive system.

The two prominent non-radiological methods have been based on the use of

electromagnetic trackers and on shape estimation from curvature measurements.

Systems based on electromagnetic tracking use two components – a field generator and

a magnetic field sensor [12] [13] [14]. The field generator provides a working volume

and the electromagnetic sensor placed within this volume generates an induced voltage


Instruments

87

dependent on the excitation field strength at that location. These systems can provide

high accuracy. However, they are affected by the presence of other ferromagnetic

objects in the operating environment. The accuracy is also dependent on the distance of

the sensor from the field generator. Placing the field generator in the vicinity of the

patient in an already cramped operating environment is a challenge.

The other prominent non-radiological methods are based on estimating the curvature of

the scope. Different sensing modalities including Micro Electro-Mechanical Sensors

[15] [16] and FBG sensors have been use for this purpose. FBG sensors in particular

have become popular for curvature estimation. FBG sensors fabricated in single mode

fibers have mostly been used for shape estimation of flexible solid needles made of

nitinol (a nickel-titanium alloy) [17] [18]. Though a method based on FBG sensors for

endoscope shape estimation has been proposed in earlier work [19], it does not provide

details of measured radii of curvatures and curve reconstruction. Other alternatives to

FBGs in single mode fibers have been recently proposed. These include FBG sensors

fabricated in multi-core fibers [20] and FBG sensors fabricated in no-core fibers using

direct femtosecond laser inscription [21]. These alternatives are interesting. However,

these require complex setups and specialty fibers which make use of standard sensor

fabrication and interrogation techniques more difficult.

One of the main advantages of FBG sensors which differentiate them from other sensors

is the ability to embed them in other materials during the curing process which can be

done at elevated temperatures. In this work, a novel method based on FBG sensors

fabricated in single mode fibers embedded inside a polymer filled tube has been

proposed to estimate the shape based on curvature.


Instruments

88

4.4 Shape Estimation from curvature using beam

theory

Beam theory can be used to determine the curvature of a beam under deflection using

measurements of the strains induced at the surface of the beam due to the deflection as

shown in figure 4.5

Figure 4.5 Beam deflection

Beam theory makes the following assumptions

Plane cross sections, normal to the neutral axis remain plane during bending (no

shear)

Hook’s law is applicable and Young’s modulus for tension and compression are

same

The beam is initially straight and all longitudinal fibers bend into concentric

circular arcs, with radii of curvature that are much larger compared to the cross

sectional dimensions of the beam

Neutral Axis

c d

e f y

R


Instruments

89

Under these assumptions, the strain along the small segment ‘ef’ can be calculated as

𝜀𝑥 = 𝑒𝑓−𝑐𝑑

𝑐𝑑

= (𝑅−𝑦)∆𝜃−𝑅∆𝜃

𝑅∆𝜃= −

𝑦

𝑅 --- (4.1)

Using equation 4.1, and knowing y and the measured strain, it is possible to estimate

the radius of curvature R.

4.5 Strain estimation using FBG sensors bonded on

endoscope and on a nitinol wire

Using the simple bending theory, an experiment has been set up to determine if direct

estimation of radius of curvature is possible from strain measurements using FBG

sensors bonded directly on the endoscope as shown in figure 4.6.

Figure 4.6 Setup for curvature estimation using FBG sensor bonded on the endoscope


Instruments

90

The estimated radius of curvature and the errors are listed in table 4.2. It can be clearly

seen that the errors are almost an order of magnitude higher than the actual radius of

curvature. This can be attributed to the geometry of the endoscope which consists of a

combination of multiple channels, the protective sheath on the endoscope and the

presence of pull wires inside the endoscope.

Table 4.2 Estimated Radius of curvature using strains from FBG sensor bonded on the

endoscope

Actual

Radius of

Curvature

(mm)

Measured strain

(μstrain)

Estimated Radius

of Curvature using

y = 5mm

(mm)

Error

(mm)

164mm 3600 1404 -1240

130mm 4000 1247 -1117

96mm 4900 1012 -916

To overcome the problem, an option is to use a nitinol wire on which FBG sensors can

be bonded and this can be inserted into the working channel of the endoscope. Nitinol

is a nickel-titanium alloy with an elastic modulus of approximately 40 GPa. It is

significantly more flexible compared to stainless steel which has an elastic modulus of

193 GPa. It has been used for medical components like dental implants and stents. The

experimental setup used before has been modified to use a nitinol wire with 1.2mm

diameter on which FBG sensor has been bonded as shown in figure 4.7 and the strains

have been measured.


Instruments

91

Figure 4.7 Experimental setup with FBG sensor bonded to a nitinol wire

Table 4.3 Results of curvature measurements using nitinol wire

Actual Radius of

Curvature

(mm)

Measured

strain

(μstrain)

Calculated Radius of

Curvature using y =

0.605mm

(mm)

Error %

164 3800 158 3.8

154 4100 147 4.7

141 4600 130 8.4

130 4800 125 4

120 5100 118 1.6

108 5800 103 4.8

96 6300 95 1.05


Instruments

92

The results in table 4.3 show that using nitinol wire is feasible for estimation of radius

of curvature. The small errors can be eliminated though calibrations. A three sensor

configuration as used in needle deflection estimation in [17] [18] to handle rotational

invariance can be extended for this application. However, the strains are very high when

the radius of curvature is below 100 mm. This creates three problems – spectrum

distortion, need for large bandwidth allocation to each sensor restricting the extension

of this method to large number of sensing points and frequent sensor breakage.

Even though higher mechanical sensor strength can be obtained for sensors which are

fabricated using draw tower process and immediate recoating, it is still difficult to

achieve radius of curvatures as small as 30mm. The spectrum distortion leads to

difficulty with interrogation as shown in figure 4.8.

(a)

(b)

Figure 4.8 (a) spectrum for unstrained sensor and (b) spectrum for sensor at strain of

6000µstrains during bending

Even for interrogation systems using swept tunable lasers with narrow line-width, this

is a problem as multiple peaks can occur. Such distortion excludes the use of lower cost

interrogation systems based on linear detector arrays and curve fitting using Gaussian

spectrum approximation which can potentially be used when the number of sensors per

fiber is small. During endoscopy it is easily possible to have curvatures up to 30mm for

shapes of endoscope as shown in figure 4.9.


Instruments

93

Figure 4.9 radius of curvature for bends during endoscopy

In order to demonstrate the effect of putting the nitinol wire in the endoscope channel

on the endoscope’s flexibility, forces required for same amount of bending has been

measured. The endoscope has been fixed at one end and the other end has been put in a

fixture with a force sensor. The fixture has been linearly translated and the experiment

has been repeated for two bending positions with and without the nitinol wire inside the

endoscope as shown in figure 4.10.

Figure 4.10 Setup to qualitatively evaluate the impact of the nitinol wire put inside the

endoscope on the flexibility of the endoscope. The left end is fixed while the right end

is translated to change the bending radius of the endoscope. A force sensor on the right

side measures the force at the tip of the endoscope.


Instruments

94

In the measurement, the forces corresponding to bending as seen in figure 4.10 (a) are

1.62 N and 3.02N without and with the nitinol wire respectively. The forces

corresponding to bending as seen in figure 4.10 (b) are 1.95N and 4.02N without and

with the nitinol wire respectively. Thus, putting the nitinol wire inside the scope almost

doubles the force required to achieve the same amount of bending. This significant

adverse effect on the flexibility of the endoscope is undesirable and hence and alternate

approach is required.

4.6 Shape reconstruction from curvature

The shape can be reconstructed from a series of curvature measurements by considering

it to be a curve parameterized by its length ‘s’. When the tip with the sensors moves

along this curve, it measures different strains depending on the curvature and this can

be used to reconstruct the curve as shown in figure 4.11.

Figure 4.11 Estimating shape using curvatures measured by a sensing point close to

the tip moving along the structure


Instruments

95

The Frenet-Serret equations can be written as

�̇�(𝑠) = 𝜅(𝑠) 𝑵(𝑠)

�̇�(𝑠) = −𝜅(𝑠) 𝑻(𝑠) − 𝜏(𝑠) 𝑩(𝑠)

�̇�(𝑠) = 𝜏(𝑠) 𝑵(𝑠)

where T(s) is the unit tangent, N(s) is the unit normal, B(s) is unit binornal, 𝜅(𝑠) is the

curvature at ‘s’ and 𝜏(𝑠) is the torsion at ‘s’. The points on the curve at which

measurements are made can be considered to be equally spaced.

For in-plane bending, 𝜏(𝑠) is zero. Hence the equations can be simplified and written in

matrix notation as

[𝑻(𝑠)̇

𝑵(𝑠)̇] = [

0 𝜅(𝑠)

−𝜅(𝑠) 0] [

𝑻(𝑠)

𝑵(𝑠)] --- (4.2)

Given the starting values T(0), N(0) and 𝜅(0), and strain measurements at regular arc

length of ‘l’, the curvatures 𝜅(1), 𝜅(2),… can be estimated and hence the curve can be

reconstructed using small approximate circular segments with arc lengths l [22] as

shown in figure 4.12.

Given starting coordinate x(0) and T(0), N(0), 𝜅(𝑠) and l

xh(0) = x(0)

T1 = T(0)

N0 = N(0)


Instruments

96

The centre of the circular segment can be obtained as

Cj = 𝒙𝒉(𝑠𝑗) + (1

𝜅(𝑠𝑗)) 𝑵𝑗 --- (4.3)

𝒙𝒉(𝑠𝑗) = 𝒄𝒋 +1

𝜅(𝑠𝑗)[−𝑐𝑜𝑠(𝜅(𝑠𝑗)𝑙)𝑵𝒋 + sin (𝜅(𝑠𝑗)𝑙)𝑻𝑗] --- (4.4)

Figure 4.12 Reconstruction using approximate circular segments

T and N can be updated as

𝑻𝑗+1 = 𝑠𝑖𝑛(𝜅(𝑠𝑗)𝑙)𝑵𝑗 + 𝑐𝑜𝑠(𝜅(𝑠𝑗)𝑙)𝑻𝑗 --- (4.5)

𝑵𝑗+1 = 𝑐𝑜𝑠(𝜅(𝑠𝑗)𝑙)𝑵𝑗 − 𝑠𝑖𝑛(𝜅(𝑠𝑗)𝑙)𝑻𝑗 --- (4.6)

One of the challenges in using this method is that it cannot deal with the change in

direction of the curve after a point where the curvature become zero i.e. where the curve

is almost a straight line. However, when FBG sensors are used, the direction of

curvature is also known as the measured strains also change their sign. Hence the sign

of the normal N can be updated using this information in the software implementation.

xh(s0)

N(s0)

xh(s1)

N(s1)

C0

xh(s2)

N(s2)

C1

)(/1 0s

)(/1 1s

T(s0)

T(s1)

T(s2)


Instruments

97

4.7 Curvature estimation using FBG sensors embedded

in a PDMS filled highly flexible tube

Since FBG sensors can be embedded inside other materials during the curing process,

an alternate device configuration has been created in this work for curvature estimation.

FBG sensors fabricated in a single mode fiber have been embedded in

Polydimethylsiloxane (PDMS) inside a plastic tube as shown in figure 4.13. PDMS has

been chosen because it is biocompatible. Also, it has a very low elastic modulus. The

elastic modulus when base elastomer to curing agent ratio of 10:1 is used is in the range

of 1-9MPa depending on curing temperatures [23]. This makes the sensing tube highly

flexible. The outer diameter of the tube is ~2.1mm and the inner diameter is ~1.2mm.

The fibers are placed under slight strain to avoid bending during the curing process.

After sensors have been placed in the tube which is rigidly held, PDMS with a base to

curing agent ratio of 10:1 has been injected into the tube. Care has been taken that no

bubbles are formed during the injection of PDMS. This has been cured at 70°C for 1

hour. The curing temperature has been restricted to this value to avoid distortion of the

plastic tube shape during curing. The resulting tube is highly flexible and can be bent

up to very small radii of curvature up to 3cm without breaking the sensors.

Figure 4.13 Cross section of the shape sensing tube

Plastic Tube

PDMS

(Base Polymer : Curing

Agent = 10:1)

FBG

Sensors


Instruments

98

The rotation of the tube is another parameter that has to be accounted for. In order to

understand and calibrate the response of the two FBG sensors for different rotations and

radii of curvature, the shape sensing tube has been inserted inside a tube as shown in

figure 4.14. The outer tube has been bent to have different radii of curvature between

3cm and 15cm. Measurements have been taken for six equally spaced rotation angles of

the sensing tube.

Figure 4.14 Calibration setup – the inside tube is the sensing tube with FBG sensors

Estimation of radius of curvature depends on the feasibility of the tube to be used for a

large radius of curvature of 15cm to a small radius of curvature of 3cm for different

rotational position of the sensors with respect to the neutral axis. Figure 4.15 shows the

wavelength shifts for different points during the calibrations. The wavelength shifts are

sufficiently detectable even though the response is not linear for the entire range. The

maximum shift is between +2nm and -2nm. Thus, if the system has to be extended for

multiple sensing points along the tube, the wavelength separation of the sensors has to

be more than 4 nm. The non-linear response requires the estimation of a function to map

the wavelength shifts to the radius of curvature.


Instruments

99

Figure 4.15 Examples of wavelength shifts for varying radii of curvature and

rotational position


Instruments

100

Figure 4.16 Plot showing wavelength shifts for different radii of curvature

In order to estimate the radius of curvature ‘R’ from a measured pair of wavelength shift

for each sensor, it is required to estimate the mapping function corresponding to figure

4.16. Artificial neural networks are good at learning such functions. An Artificial Neural

Network (ANN) in a configuration to perform non-linear regression has been used for

this purpose as shown in figure 4.17. A logistic [24] activation function has been used

for the neural-net nodes in the hidden layers. The two input nodes correspond to the

wavelength shifts corresponding to the bending strain. The output node provides the

radius of curvature. The ‘neuralnet’ package [25] in the statistical analysis software ‘R’

[26] has been used for training the neural network. The training has been performed

using back propagation.


Instruments

101

Figure 4.17 An artificial neural network to perform regression for estimation of radius

of curvature. The shown ANN shows the structure and weights after training based on

data presented in figure 4.16

Generally, for methods based on machine learning, the data is split as 60% for training

and 40% testing. However, having a smaller training set and larger test set ensures better

generalization. For the data generated in this experiment, only 30% of data was used for

training while the rest 70% was used as test set to ensure good generalization over the

different positions. To get reproducible results, the seed for the pseudo random number

generator has been fixed. The results of the tests are shown in figure 4.18. It can be seen

that in general, the median errors are within +/- 10% and on an average are within +/-

5%. The higher errors at some points are due to a few measurement points having high

errors in measured wavelength shifts. This is more for R=15cm as the wavelength shifts

are small and any small measurement error in wavelength shifts leads to large errors in

overall measurements. This is due to manual handling of the shape sensing tube during

the calibration. This method can be improved by using a more controlled calibration

setup and thus seems promising for determination of radius of curvature.


Instruments

102

Figure 4.18 Error plot for tests using the trained neural network

In order to test the accuracy of estimated radii of curvature and the effect of inaccuracies

on the reconstructed shape, the shape sensing tube has been moved along a short sample

curve with each measurement point 0.5mm away corresponding to l=0.5 in equations

4.4 to 4.6. The original shape has been captured using a graph sheet with 1mm markings.

The estimated radii of curvature ‘R’ for each point has been estimated using the trained

ANN shown in figure 4.17. The reconstructed curve has been plotted using the estimated

values of ‘R’ iteratively in equations 4.3 to 4.6 which have been implemented in Python.

Figure 4.18 shows the original and reconstructed curve. Even though the errors are not

small enough for use in exact localization of the tip, the method can be used for

estimation of shape traced by the tip for visualization purposes to assist the surgeons.


Instruments

103

Figure 4.19 Reconstruction of a sample small curve using measured wavelength shifts

and estimated radii of curvature using the ANN

4.8 Summary

The different methods for shape estimation of flexible medical instrument like

endoscopes that have been tried out have been presented in this chapter. Issues in using

FBG sensors bonded on nitinol have been discussed and an alternative methods based

on FBG sensors embedded inside a PDMS filled tube has been presented. This enables

shape estimation for highly flexible instruments up to very small radii of curvature

without having an adverse effect on the flexibility of the instrument. The methods

demonstrated here can determine shapes in a plane. In the current system one sensing

point close to the tip has been used. This is capable of tracing the shape during insertion.

In future, the system can be extended to have multiple sensing points so that better

interpolation can be used for estimating the shape followed during insertion. Handling

of out of plane bending to determine complete 3D shape estimation seems possible and

needs to be studied further.


Instruments

104

References

[1] S. Varadarajulu, S. Banerjee, B. A. Barth, D. J. Desilets, V. Kaul, S. R. Kethu, M.

C. Pedrosa, P. R. Pfau, J. L. Tokar, A. Wang and others, "GI endoscopes,"

Gastrointestinal endoscopy, vol. 74, pp. 1-6, 2011.

[2] "Endoscopic Procedures," 2014.

[3] C. G. L. Cao and P. Milgram, "Disorientation in minimal access surgery: A case

study," in Proceedings of the Human Factors and Ergonomics Society

Annual Meeting, 2000.

[4] J. D. Waye, "Difficult colonoscopy," Gastroenterology \& hepatology, vol. 9, p.

676, 2013.

[5] A. Parra-Blanco, M. R. Arnau, D. Nicolás-Pérez, A. Z. Gimeno-Garc\́ia, N.

González, J. A. D\́iaz-Acosta, A. Jiménez and E. Quintero, "Endoscopic

submucosal dissection training with pig models in a Western country," World

J Gastroenterol, vol. 16, pp. 2895-2900, 2010.

[6] I. A. Mouzas, "Stimulation through simulation. The use of simulators in

gastrointestinal endoscopy training," ANNALS OF

GASTROENTEROLOGY., vol. 17, pp. 23-25, 2004.


Instruments

105

[7] E. Samur, L. Flaction and H. Bleuler, "Design and evaluation of a novel haptic

interface for endoscopic simulation," IEEE Transactions on Haptics, vol. 5,

pp. 301-311, 2012.

[8] S. Chakravarthy, K. Avinash, G. Ramu and G. Ananthasuresh, "Design of an

Endoscopic Haptic Display System using an Integrated Ring-actuator," in 1st

International and 16th National Conference on Machines and Mechanisms

(iNaCoMM2013), 2013.

[9] K. Triantafyllou, L. D. Lazaridis and G. D. Dimitriadis, "Virtual reality simulators

for gastrointestinal endoscopy training," World journal of gastrointestinal

endoscopy, vol. 6, p. 6, 2014.

[10] S. Chakravarthy and G. K. Ananthasuresh,

https://mimykcom.wordpress.com/about/.

[11] D. G. Adler, "The role of fluoroscopy in the endoscopic management of luminal

gastrointestinal disorders," Techniques in Gastrointestinal Endoscopy, vol.

9, pp. 189-194, 2007.

[12] H. Atsumi, M. Matsumae, A. Hirayama, H. SHIGEMATSU, G. INOUE, J.

NISHIYAMA, M. YOSHIYAMA and J. TOMINAGA, "Newly developed

electromagnetic tracked flexible neuroendoscope," Neurologia medico-

chirurgica, vol. 51, pp. 611-616, 2011.


Instruments

106

[13] T. Reichl, J. Gardiazabal and N. Navab, "Electromagnetic servoing—A new

tracking paradigm," IEEE transactions on medical imaging, vol. 32, pp.

1526-1535, 2013.

[14] Z. Yaniv, E. Wilson, D. Lindisch and K. Cleary, "Electromagnetic tracking in the

clinical environment," Medical physics, vol. 36, pp. 876-892, 2009.

[15] T. N. Kulatunga, R. A. G. P. Ranasinghe, R. A. C. Ranathunga, R. A. C. H.

Ratnayake and N. D. Nanayakkara, "Real time endoscope trajectory tracking

in the 3D space using MEMS sensors," in Industrial and Information Systems

(ICIIS), 2013 8th IEEE International Conference on, 2013.

[16] Z. Zhang, J. Shang, C. Seneci and G.-Z. Yang, "Snake robot shape sensing using

micro-inertial sensors," in Intelligent Robots and Systems (IROS), 2013

IEEE/RSJ International Conference on, 2013.





[18] R. J. Roesthuis, M. Kemp, J. van den Dobbelsteen and John and S. Misra, "Three-

dimensional needle shape reconstruction using an array of fiber bragg grating

sensors," IEEE/ASME Transactions on mechatronics, vol. 19, pp. 1115-

1126, 2014.


Instruments

107

[19] L. Zhang, J. Qian, Y. Zhang and L. Shen, "On SDM/WDM FBG sensor net for

shape detection of endoscope," in Mechatronics and Automation, 2005 IEEE

International Conference, 2005.

[20] J. P. Moore and M. D. Rogge, "Shape sensing using multi-core fiber optic cable

and parametric curve solutions," Optics Express, vol. 20, pp. 2967-2973,

2012.

[21] K. K. C. Lee, A. Mariampillai, M. Haque, B. A. Standish, V. X. D. Yang and P.

R. Herman, "Temperature-compensated fiber-optic 3D shape sensor based

on femtosecond laser direct-written Bragg grating waveguides," Optics

express, vol. 21, pp. 24076-24086, 2013.

[22] E. Carlen, Calculus++: Description and prediction of motion, 2013.

[23] I. D. Johnston, D. K. McCluskey, C. K. L. Tan and M. C. Tracey, "Mechanical

characterization of bulk Sylgard 184 for microfluidics and

microengineering," Journal of Micromechanics and Microengineering, vol.

24, no. 3, p. 035017, 2014.

[24] C. M. Bishop, Pattern recognition and machine learning, springer, 2006.

[25] S. Fritsch, F. Guenther, M. Suling and S. M. Mueller, Training of Neural

Networks, CRAN, 2016.

[26] The R Foundation, The R Project for Statistical Computing https://www.r-

project.org/.


Instruments

108

Chapter 5 Evaluation of FBG interrogation systems

based on InGaAs linear detector arrays and curve fitting

methods using Gaussian approximation

109

Chapter 5

Evaluation of FBG interrogation

systems based on InGaAs linear

detector arrays and curve fitting

methods using Gaussian

approximation

5.1 Introduction

FBG interrogation systems provide the mechanism to measure the Bragg wavelength

shifts. For any application involving FBG sensors, interrogation systems are as

important as sensor characteristics [1]. Successful and widespread adoption of FBG

sensors in biomedical applications is dependent on the availability of cost-effective

interrogation systems meeting the application specific requirements on accuracy and

speed and the possibility of integration with existing measurement systems for real-

world use [2].

Several FBG interrogation systems have been proposed in literature. One of the

simplest, earliest and widely adapted method is based on the use of an edge filter [3] [4]

[5] [6] providing a linear relationship between the wavelength shifts and the output

intensity changes of the filter as shown in figure 5.1.




110

Figure 5.1 Schematic diagram of interrogation system using edge filter [3]

The filter can be modeled as

F(λ) = A(λ – λ0)

where A is the filtering slope and F(λ) is zero at λ0.

Assuming back reflected Bragg peak follows a Gaussian profile with center wavelength

λb and spectral width Δλ, the intensity of the reference and the filtered beams can be

represented as

𝐼𝑟𝑒𝑓 = 𝐼0 𝑅 √𝜋

2∆𝜆

𝐼𝑓𝑖𝑙𝑡 = 𝐼0 𝑅 𝐴 √𝜋

2(𝜆𝑏 − 𝜆0 +

Δ𝜆

√𝜋) ∆𝜆

where I0 is the intensity of the light incident on the grating and R is the grating

reflectivity.

Source

Linear Edge Filter F(λ) = A(λ – λ0)

Photodetector Photodetector

Iref Ifilt

Electronics

FBG Sensor




111

The ratio of the two outputs is

𝐼𝑓𝑖𝑙𝑡

𝐼𝑟𝑒𝑓= 𝐴 (𝜆𝑏 − 𝜆0 +

Δ𝜆

√𝜋)

The ratio varies linearly as the wavelength λb shifts and is independent of the source

intensity fluctuations. The method has been used with different filtering elements and

even all-fiber approaches have been used to avoid limitations of bulk optic filters. The

technique is cost effective and accuracies can also be improved considering application

specific calibration. However, extending it for use with multiple sensors is difficult as

simultaneous use of several edge filtering components and couplers is not practical.

(a) (b)

Figure 5.2 (a) Schematic of matched FBGs based interrogation and (b) simultaneous

multiple sensor interrogation [7] [8]

A natural simple alternative to this method is to use methods based on matched FBGs

[7] [8] [9] [10] [11] [12] [13] [14]. In its simple form, the concept consists of using a

grating pair of sensing grating (Gs) and receiving grating (Gr) as shown in figure 5.2(a).

The center peak wavelength of both are identical when both experience the same strain

Source G

G

Piezo-stretcher

Photodetector Controller

Source Gs Gs

Gr Gr

Piezo-stretcher




112

and temperature. The back-reflected light from Gs is passed through a coupler to Gr

which is mounted on a controlled piezo-stretcher. The back-reflected light is passed to

a photodetector and the intensity is observed which is maximum when the peak

wavelengths match. The peak wavelength of the sensing grating Gs varies in response

to strain or temperature changes. A sweep signal is applied to the piezo-stretcher. The

intensity of back reflected light from Gr is dependent on the matching condition. The

stretching leads to change in intensity at the photodetector.

These methods provide a cost effective mechanism for FBG interrogation. However, it

is difficult to use these for real world applications due to the constraint of using very

closely matched FBGs in both the sensing arm and the receiving arm. Also, the number

of sensors that can be interrogated is pre-determined which is a major constraint.

Figure 5.3 Schematic diagram of interrogation system using tunable fabry-perot filter

[15] [16]

Another interrogation scheme for which several extensions and modifications have been

proposed over the years [15] [16] [17] [18] [19] consists of using a tunable fabry-perot

filter as shown in figure 5.3.

Broadband Source

Fabry-perot filter ∑

Feedback Control

Detector




113

The fabry-perot passband is locked to the back reflected signal from the FBG sensor

using a feedback loop. For a fiber fabry perot filter, the tuning can be performed using

piezoelectric adjustment of the cavity spacing at a modulating frequency fm. The output

contains components of the fundamental and harmonics of fm. When the wavelengths of

the fabry-perot filter and the back reflected signal are aligned, the amplitude of the

fundamental is nulled. The tuning constrains the sampling rate and the accuracy of the

system. In recent times high accuracies have been achieved. However, cost effectiveness

still remains a constraint.

Interrogation systems based on narrow linewidth wavelength tunable sources have been

popular in recent years. Different configurations have been used for source. These

include wavelength swept fiber lasers with fabry-perot filter in the cavity to sweep the

laser output wavelength in time [20], wavelength swept fiber lasers with intracavity

acoustooptic tunable filter [21], swept fiber laser based on dispersion tuning technique

which does not require any optical tunable filter in laser cavity [22], Fourier domain

mode locking laser [23] [24] and active mode locking fiber laser [25]. Interrogation

using tunable swept lasers can provide high accuracy. Availability of a high power

budget per sensor enables interrogation of large number of sensors over long distances.

However, at present the cost of swept tunable laser based systems is high. This is

acceptable for applications which require monitoring of large number of sensors

distributed over large distances and also for experimental setups and niche applications.

However, for biomedical applications involving small number of sensor per setup, the

cost is an important factor. The expectation is that technology advancements might

bring down the cost in future.

For several biomedical and robotic applications, only a few FBG sensors per fiber are

required. Meeting application specific requirements for accuracy and resolution along

with required sampling rates and the feasibility of creating systems with integrated




114

embedded hardware for real-time computations are sufficient to make widespread

adoption easier. Interrogation systems based on broadband source and InGaAs linear

detector arrays provide an option. Figure 5.4 shows the schematic diagram of an

interrogation system based on transmission grating and linear detector array.

Figure 5.4 Schematic diagram of an interrogation system with InGaAs linear detector

array [26]

The main challenge in using such interrogation systems is achieving the required

resolution. For example, an InGaAs linear photodiode array with 512 pixels with a

wavelength range of 1510nm to 1595 nm provides an FWHM resolution of

approximately 330 pm. This is not sufficient for most applications but curve-fitting

methods [27] [28] have been used to achieve sub-pixel accuracy. However, the effect of

choice of number of pixels around a spectrum peak and the choice of algorithms has a

significant impact on the accuracy and computational efficiency of the curve fitting

algorithm. In this work, a detailed analysis of the effect of the choice of number of pixels

in the region of the peak on the accuracy of the estimated Bragg wavelength and the




115

points where errors are highest is provided. This is important because specifications

related to minimum errors are not sufficient for determining usage in several

applications. It is important to choose the algorithms based on the knowledge of

average and maximum errors and the conditions which result in maximum errors.

It is also useful to implement the algorithm on embedded hardware and hence it is

important to consider computational complexity to meet the sampling rate

requirements. Hence experiment has been designed to be able to detect the

maximum errors in the system for multiple sensors and all algorithms have been

evaluated using the data obtained from these measurements.

5.2 Measurement Setup and Data Characteristics

Two commercially available interrogation systems – I-MON 512 USB from Ibsen

Photonics and SM130 from Micron Optics have been used for the work presented in

this thesis. The SM130 is based on a swept tunable laser while the I-MON 512 USB

uses a transmission grating along with an InGaAs linear detector array with 512 pixels.

I-MON 512 has been selected because it allows direct access to the measured intensity

at each pixel and the choice of algorithm for peak estimation is left to the user. The

focus of this work is on the comparison of algorithms and they would be applicable for

any similar system.

5.2.1 Setup

Most sensing applications using FBG sensors are based on measuring the wavelength

shift in response to strain or temperature changes. Hence, for the analysis of the

algorithms, six FBG sensors have been fabricated on a single germanosilicate fiber

using phase mask technique and small incremental strains have been applied using the

setup as shown in figure 5.5. The sensors have been fabricated using three different




116

phase masks. For each phase mask, two sensors have been fabricated close to each other

with a separation of approximately 15mm by pre-stretching the fiber and translating it.

The wavelength separation between these two sensors is approximately 1.5nm.

Figure 5.5 Setup to generate data samples

The fiber has been fixed rigidly to both the translation stage and the fixed end. Not

having the fiber bonded to another material avoids any non-linearity or spectrum

distortion arising because of the bonding. Since the strain on each sensor has been

applied thorugh elongation by using the translation stage, and measurements for each

sensor have been made simultaneously, the effect of slight variations in ambient

temperature can be neglected as the temperature variation for all sensors is similar.

5.2.2 Data Characteristics

The translation stage has been linearly moved for 10 steps of 10µm each. At each step,

the data has been acquired using both the interrogation systems and has been logged in

files. The log of data from I-MON 512 consists of the complete spectrum across all

pixels. The log of data from the SM130 consists of the intensity measured at 512 pixels.

There are small variations in the applied strain at each measurement step because of

small inaccuracies in the linear translation stage. However, the variation is not relevant




117

for this evaluation as error comparisons have been made for each measurement point.

Figure 5.7 shows the spectrum obtained from the linear detector array for one sample

measurement. In the plot, a linear interpolation has been added between the points to

make visualization of peaks easier.

Figure 5.6 Spectrum obtained from the linear detector array

5.3 Bragg Wavelength Estimation

The linear detector array only provides the intensity at each pixel. Using the pixel with

highest intensity as the one representing peak wavelength would lead to very low

resolution. Hence the peak wavelength has to be estimated using an approximation

algorithm providing sub-pixel resolution.




118

5.3.1 Centroid Algorithm

The simplest method to achieve sub-pixel resolution is based on the centroid algorithm

[29] as shown in equation 5.1

𝜆𝑏 =∑ 𝜆𝑖𝐼𝑖

𝑁𝑖=1

∑ 𝜆𝑖𝑁𝑖=1

--- (5.1)

where 𝜆𝑖 are the wavelengths at the pixels around the peak and 𝐼𝑖 are the corresponding

intensity values.

The algorithm has been applied to estimate wavelength shifts for all sensors for the all

the measurements corresponding to incremental strains. The main advantage of the

algorithm is its low computational cost. As is evident from equation 5.1, only a small

number of multiplications and additions and one division are required. The computation

time does not constrain the sampling rates for the overall system even when combined

with a visualization system. The computed values of Bragg wavelength shift using the

centroid algorithm along with that measured using the swept tunable laser based

interrogation system are shown in figure 5.7.

The graphs show the advantage of using the sensor configuration and small incremental

strains for the analysis. For each sensor, there is a corresponding measurement where

the error in measured wavelength shift is very high relative to other measurements. The

absolute value of change in strains for each sensor are not important for this analysis as

only the error in estimation of the same wavelength shift using different techniques is

evaluated.




119

Figure 5.7 Errors in peak wavelength estimation using the centroid method compared

to measurement using the interrogation system based on swept tunable laser

The spectrum shift at the points where the deviation is maximum has been analyzed. It

has been found that the deviation is maximum when the spectrum shifts to include a

new pixel. This can be seen in the comparison as shown in figure 5.8.

For wavelength shift due to strain difference between measurement 0 and measurement

1, the spectrum for sensor 1 shifts across pixels. This does not happen for the other

sensors. Similarly, for wavelength shift between measurement 1 and measurement 2,

the spectrum shifts across pixels for sensor 2. The errors in estimation are highest at

these points.




120

Figure 5.8 Plots showing the points around the peak. Each measurement corresponds

to a shift in wavelength due to applied strain. The original and the shifted points and

the estimated peak using centroid algorithm are shown for three different sensors on

the fiber.

5.3.2 Gaussian Approximation for the FBG spectrum

and least square estimation of Bragg

wavelength

The FBG spectrum can be approximated as a Gaussian [30] as represented by equation

5.2.

𝐼(𝜆, 𝐴, 𝜆𝑏 , 𝜎)= 𝐴𝑒−

(𝜆−𝜆𝑏)2

2𝜎2 --- (5.2)

where 𝜆𝑏represents the Bragg wavelength.




121

Given the wavelengths at the pixels (𝜆𝑖)and the intensities (Ii), from the measurements,

the goal is to estimate the parameters𝜆𝑏, A and 𝜎. This can be done using least squares

estimation. The error function is given by

𝐸 = ∑ (𝐼(𝜆𝑖, 𝐴, 𝜆𝑏 , 𝜎) − 𝐼𝑖)2𝑁𝑖=1 --- (5.3)

has to be minimized with respect to the parameters 𝜆𝑏 , A and 𝜎. Though there are

several methods that can be used to solve this, the Levenberg-Marquard algorithm [31]

[32] has been extensively used and is the default implementation in most libraries (e.g.

the least squares curve fitting function in LabView uses it). In this evaluation, the

implementation in the SciPy [33] library in Python which internally uses the MINPACK

[34] implementation has been used.

To apply, the non-linear least square estimation, it is first required to find the pixels in

the neighborhood of peaks. For extracting these pixels from the data for all the 512

pixels in the spectrum, a simple peak search algorithm has been employed. This first

applies a threshold based on the intensity of the highest intensity peak in the spectrum

to get all possible peak regions. Then the change in gradient of the difference in

intensities has been used to detect all the pixels with highest intensities in a

neighborhood. This method enables handling of sensors with varying reflectivity. The

need for this is evident by observing the spectrum in figure 5.6 where the smallest peak

has an intensity approximately 50% of that of the highest peak. Once all the pixels with

highest intensities has been obtained, the least square method can be employed based

on different number of pixels around the pixel with maximum intensity.

The Levenberg-Marquardt algorithm requires setting of initial values for the parameters

which are being estimated. The wavelength of the pixel with highest intensity and the

intensity at that pixel have been used as the initial values for 𝜆𝑏 and A respectively.




122

Figure 5.9 Estimated Bragg wavelengths using different number of pixels around the

peak




123

To evaluate the variations in the estimated peak wavelength, the least square estimation

has been performed using four different set of pixels. In the following discussion, peak

refers to the pixel with highest intensity. The first variation uses one pixels to the left

and one to the right of peak, the second uses four pixels with two pixels to the left of

the peak, the third uses four pixels with two pixels to the right of the peak and the fourth

uses five pixels with two pixels on both sides of the peak. Figure 5.9 shows the estimated

wavelengths for all sensors for all measurements.

Since accuracy in measurement of wavelength shift is important, it is analyzed for all

sensors for each measurement point. The comparisons have also been made to the

wavelength shift measured by the interrogator based on swept tunable laser. It has been

found that the accuracy varies with the different number of pixels used in the non-linear

least squares estimation as shown in figure 5.10.

Figure 5.10 Wavelength shift measured using different number of pixels and

compared to measurement with swept tunable lased based interrogation system.




124

To find the main source of high variation at certain points, the spectrum shift at these

points has been analyzed. It has been found that similar to the centroid algorithm, the

maximum variation occurs when the spectrum shifts completely one pixel to the left or

right. This is seen in figure 5.11. For the purpose of curve fitting, pixels are assumed as

points (representing the center of the pixel) and the intensities as the values at these

points. In reality, the intensity is spread across the pixels. Hence, when the spectrum

shift corresponding to wavelength shift from one measurement to the next is not limited

to the same set of pixels, the error is higher. Unlike in the case of centroid algorithm,

using five pixels instead of three does not improve the accuracy. It actually reduces it

as the curve fitting works best only when the pixels are within the Full Width at Half

Maximum (FWHM) range.

Figure 5.11 Spectrum for sensor 1 and sensor 2 before and after wavelength shift due

to applied strain

Since all measurements have been done by applying increasing strain, the wavelength

shift is in the same direction. This is direction towards left in terms of pixels in the linear




125

detector array as pixel number one represents the maximum wavelength limit of

1595nm. Hence, at the error points, the impact is highest on the peak estimation using

four points where the two points are to the left of the pixel with highest intensity as there

is a high asymmetry around the peak after the shift.

Figure 5.11 clearly shows that the variation between the estimated wavelength shifts is

very small for the different number of pixels for sensor 2 because the spectrum does not

shift across the pixels. However, the estimated wavelength shift vary significantly for

sensor 1 because of the shift in spectrum by one pixel to the left. Figure 5.12 shows the

plot for the difference between the estimated wavelength shifts using the Gaussian

approximation and that measured using the swept tunable laser based interrogation

system for all six sensors for the eight wavelength shift measurements. The results have

been summarized in table 5.1.

Table 5.1 Wavelength shift difference from tunable laser based system using different

number of points for non-linear least squares estimation using Levenberg Marquardt

algorithm

3

Points

4 points

(peak - 2 to

peak + 1)

4 points

(peak - 1 to

peak + 2)

5 points

Maximum wavelength

shift difference (pm)

6 32 12 14

Median wavelength shift

difference (pm)

0.3 1.7 0.8 0.8

Even though the median difference between the estimated values and those measured

using tunable laser based systems are less than 1 pm, the maximum errors are much




126

higher when using 4 points and 5 points as compared to 3 points. This is attributed to

the measurements when the spectrum shifts across the pixels.

Figure 5.12 Difference in wavelength shift estimation using Levenberg-Marquardt

algorithm and Gaussian spectrum approximation compared to measurement using

swept tunable laser based interrogation system

Table 5.2. Wavelength shift difference from tunable laser based system using

Levenberg Marquardt algorithm and Centroid Algorithm

Lev-Mar 3

Points

Centroid 3

Points

Centroid 5 points

Maximum wavelength


6 102 70

Median wavelength


0.3 14.4 9.6




127

Comparison of errors using curve fitting with Lev-Mar method and using centroid

algorithm highlights the specific advantage of using the Lev-Mar specifically to reduce

the maximum errors. The maximum error using centroid fit can be up to 102 pm using

3 points and 70 pm using 5 points while that with Lev-Mar and 3 points is limited to 6

pm. The median error is also significantly reduced from close to 15pm for centroid with

3 points to less than 1 pm for Lev-Mar with 3 points as summarized in table 5.2.

5.3.3 Lower computational cost algorithm utilizing

Gaussian Approximation for the FBG spectrum

In the previous section, it has been shown that Bragg wavelength estimation using non-

linear least square estimation and Gaussian approximation for interrogation systems

based on InGaAs linear detector arrays can achieve accuracy very close to those

achieved by interrogation systems based on swept tunable laser. The errors are lowest

when using three pixels for the least square estimation such that the center pixel is the

one with the highest intensity. However, the computational cost of non-linear least

square estimation is high in spite of using the Levenberg-Marquard algorithm. To

overcome this, Caruana’s algorithm [35] [36] has been evaluated which is specific to

curve fitting for Gaussian function and does not require any iterative method. Caruana’s

algorithm reduces the estimation to solving simultaneous linear equations which can be

done efficiently using matrix manipulation methods.

Taking natural logarithm on both sides for equation (5.2)

ln 𝐼 = ln 𝐴 −(𝜆−𝜆𝑏)2

2𝜎2 --- (5.4)

= ln 𝐴 −𝜆2

2𝜎2 +2𝜆𝑏𝜆

2𝜎2 −𝜆𝑏

2

2𝜎2 --- (5.5)

= 𝑎 + 𝑏𝜆 + 𝑐𝜆2 --- (5.6)




128

where 𝑎 = ln 𝐴 − 𝜆𝑏

2

2𝜎2 , 𝑏 = 𝜆𝑏

𝜎2 , 𝑐 = −1

2𝜎2

For measured intensities Ii at pixels with wavelengths λi, this is equivalent to fitting a

parabola to the points. Formulating it as a least squares estimation problem, the error

term to be minimized can be written as

𝐸 = ∑ (ln 𝐼𝑖 − 𝑎 − 𝑏𝜆𝑖 − 𝑐𝜆𝑖2)

2𝑁𝑖=1 --- (5.7)

Minimization can be done by solving

𝜕𝐸

𝜕𝑎= 0 --- (5.8)

𝜕𝐸

𝜕𝑏= 0 --- (5.9)

𝜕𝐸

𝜕𝑐= 0 --- (5.10)

Solving (7), (8) and (9) respectively,

∑ ln 𝐼𝑖𝑁𝑖=1 − 𝑎𝑁 − 𝑏 ∑ 𝜆𝑖

𝑁𝑖=1 − 𝑐 ∑ 𝜆𝑖

2𝑁𝑖=1 = 0 --- (5.11)

∑ 𝜆𝑖ln 𝐼𝑖𝑁𝑖=1 − 𝑎 ∑ 𝜆𝑖

𝑁𝑖=1 − 𝑏 ∑ 𝜆𝑖

2𝑁𝑖=1 − 𝑐 ∑ 𝜆𝑖

3𝑁𝑖=1 = 0 --- (5.12)

∑ 𝜆𝑖2ln 𝐼𝑖

𝑁𝑖=1 − 𝑎 ∑ 𝜆𝑖

2𝑁𝑖=1 − 𝑏 ∑ 𝜆𝑖

3𝑁𝑖=1 − 𝑐 ∑ 𝜆𝑖

4𝑁𝑖=1 = 0 --- (5.13)

[

𝑁 ∑ 𝜆𝑖𝑁𝑖=1 ∑ 𝜆𝑖

2𝑁𝑖=1

∑ 𝜆𝑖𝑁𝑖=1 ∑ 𝜆𝑖

2𝑁𝑖=1 ∑ 𝜆𝑖

3𝑁𝑖=1

∑ 𝜆𝑖2𝑁

𝑖=1 ∑ 𝜆𝑖3𝑁

𝑖=1 ∑ 𝜆𝑖4𝑁

𝑖=1

] [𝑎𝑏𝑐

] = [

∑ ln 𝐼𝑖𝑁𝑖=1

∑ 𝜆𝑖ln 𝐼𝑖𝑁𝑖=1

∑ 𝜆𝑖2ln 𝐼𝑖

𝑁𝑖=1

] ---

(5.14)

𝜆𝑏 = −𝑏

2𝑐 --- (5.15)




129

This is in the standard form Ax = b. The terms with higher powers in A can be computed

efficiently by using vector representations for data points and dot products. Since, for

the purpose of this evaluation, all implementations have been done in Python, these can

be very efficiently done. In order, to estimate the gain in efficiency, the Bragg

wavelength estimation has been done both with the non-linear least squares estimation

using Levenberg- Marquardt algorithm and Caruana’s algorithm in comparison to

Centroid algorithm. The execution time has been measured on a Raspberry Pi 2 model

B board with a 900MHz quad core ARM Cortex – A7 CPU and 1GB of RAM and

running Rasbian operating system which is based on Debian operating system. The

method for timing estimate has been done using the standard recommended method in

Python using the timeit library.

The results have been summarized in Table 5.3. As can be seen, the execution time is

significantly lower for Caruana’s algorithm as compared to Levenberg Marquardt

algorithm. It has been observed that using Caruana’s algorithm provides an average

speed up of approximately 6.5 times. Using Levenberg Marquard algorithm with the

used setup reduces the maximum sampling frequency to around 30Hz. However, using

Caruana’s algorithm allows a sampling rate around 200Hz. The estimated values for

Bragg wavelength have also been compared. The difference between Caruana’s

algorithm and Lev-Mar is less than 0.01pm which is sufficient to recommend the use of

Caruana’s algorithm over non-linear least square estimation using Levenberg-

Marquardt algorithm with three pixels on embedded hardware.

There is another advantage of using Caruana’s algorithm. When the strain changes

slowly compared to the sampling rate and the pixels for each peak do not change, there

is no need to recalculate the matrix A. This would further reduce the required

computation for each sample. However, this was not used in the computation time

evaluation here to avoid any dependency on specific data.




130

Table 5.3 Comparison of execution times and difference in estimated wavelength

from tunable laser based system for the three methods

Centroid 3

points

Lev-Mar 3

Points

Caruana 3

Points

Median Normalized

Execution time

1x 64x 10x

Maximum wavelength

shift difference from

tunable laser based

system (pm)

100 6 6

Median wavelength shift

difference from tunable

laser based system

(pm)

15 0.3 0.3

5.4 Summary

This evaluation has shown that FBG interrogation systems based on InGaAs linear

detector arrays combined with curve fitting methods utilizing Gaussian approximation

of the FBG spectrum can achieve accuracy close to that achieved by interrogation

systems based on swept tunable laser. This is useful for interrogation systems for

biomedical and robotic applications where the number of sensors and distance between

the sensors and the interrogation system is small and benefits other than high accuracy

of a swept laser baser system are not required. It has been shown that the choice of

number of pixels around the peak has a significant impact on the maximum errors when




131

using the non-linear least square estimation based on Levenberg-Marquardt algorithm.

It has also been shown that Caruana’s algorithm provides the same accuracy as

Levenberg-Marquardt algorithm while being significantly computationally faster. This

has been demonstrated through implementation on an embedded hardware platforms

which can be packaged with the linear detector array and the transmission gratings.




132

References

[1] S. D. Dyer, P. A. Williams, R. J. Espejo, J. D. Kofler and S. M. Etzel, "Key

metrology considerations for fiber Bragg grating sensors," in Smart

Structures and Materials, 2004.

[2] E. Al-Fakih, A. a. A. N. Osman and M. a. R. F. Adikan, "The use of fiber Bragg

grating sensors in biomechanics and rehabilitation applications: The state-of-

the-art and ongoing research topics," Sensors, vol. 12, pp. 12890-12926,

2012.

[3] S. M. Melle, K. Liu and R. M. Measures, "A passive wavelength demodulation

system for guided-wave Bragg grating sensors," IEEE Photonics Technology

Letters, vol. 4, pp. 516-518, 1992.

[4] T. Guo, H.-Y. Tam and J. Albert, "Chirped and tilted fiber Bragg grating edge

filter for in-fiber sensor interrogation," in CLEO: Science and Innovations,

2011.

[5] U. Tiwari, K. Thyagarajan, M. R. Shenoy and S. C. Jain, "EDF-based edge-filter

interrogation scheme for FBG sensors," IEEE Sensors Journal, vol. 13, pp.

1315-1319, 2013.

[6] N. Stan, D. C. Bailey, S. L. Chadderdon, S. Webb, M. Zikry, K. J. Peters, R. H.

Selfridge and S. M. Schultz, "Increasing dynamic range of a fibre Bragg




133

grating edge-filtering interrogator with a proportional control loop,"

Measurement Science and Technology, vol. 25, p. 065206, 2014.

[7] D. A. Jackson, L. Reekie, J. L. Archambault and A. B. L. Ribeiro, "Simple

multiplexing scheme for a fiber-optic grating sensor network," Optics

Letters, vol. 18, pp. 1192-1194, 1993.

[8] M. A. Davis and A. D. Kersey, "Matched-filter interrogation technique for fibre

Bragg grating arrays," Electronics letters, vol. 31, pp. 822-823, 1995.

[9] A. B. L. Ribeiro, L. A. Ferreira, J. L. Santos and D. A. Jackson, "Analysis of the

reflective-matched fiber Bragg grating sensing interrogation scheme,"

Applied Optics, vol. 36, pp. 934-939, 1997.

[10] A. S. Paterno, V. de Oliveira, T. S. Figueredo and H. J. Kalinowski, "Multiplexed

fiber Bragg grating interrogation system using a modulated fiber Bragg

grating and the tunable-filter method," IEEE Sensors Journal, vol. 6, pp.

1662-1668, 2006.

[11] D. Kumar, M. Raju, K. N. Madhusoodanan and others, "Implementation of

interrogation systems for fiber Bragg grating sensors," Photonic Sensors, vol.

3, p. 283, 2013.

[12] J. Cui, Y. Hu, K. Feng, J. Li and J. Tan, "FBG Interrogation Method with High

Resolution and Response Speed Based on a Reflective-Matched FBG

Scheme," Sensors, vol. 15, pp. 16516-16535, 2015.




134

[13] K. Suresha, S. Yamdagni, S. C. Mohan and S. Asokan, "A Novel Opto Electro

Mechanical Interrogation System for Measuring Wavelength Shifts in Fiber

Bragg Grating (FBG) Sensors," International Journal of Optomechatronics,

vol. 2, pp. 61-73, 2008.

[14] K. V. Madhav and S. Asokan, "Spectrum estimation by wavelength shift time-

stamping in a fiber Bragg grating sensor," IEEE Photonics Technology

Letters, vol. 16, pp. 1355-1357, 2004.

[15] A. D. Kersey, T. A. Berkoff and W. W. Morey, "Multiplexed fiber Bragg grating

strain-sensor system with a fiber Fabry--Perot wavelength filter," Optics

letters, vol. 18, pp. 1370-1372, 1993.

[16] Y. N. Ning, A. Meldrum, W. J. Shi, B. T. Meggitt, A. W. Palmer, K. T. V. Grattan

and L. Li, "Bragg grating sensing instrument using a tunable Fabry-Perot

filter to detect wavelength variations," Measurement Science and

Technology, vol. 9, p. 599, 1998.

[17] P. Tsai, F. Sun, G. Xiao, Z. Zhang, S. Rahimi and D. Ban, "A new fiber-Bragg-

grating sensor interrogation system deploying free-spectral-range-matching

scheme with high precision and fast detection rate," IEEE Photonics

Technology Letters, vol. 20, pp. 300-302, 2008.

[18] W. R. Allan, Z. W. Graham, J. R. Zayas, D. P. Roach and D. A. Horsley,

"Multiplexed fiber Bragg grating interrogation system using a




135

microelectromechanical Fabry--Perot tunable filter," IEEE Sensors Journal,

vol. 9, pp. 936-943, 2009.

[19] T. Vella, S. Chadderdon, R. Selfridge, S. Schultz, S. Webb, C. Park, K. Peters and

M. Zikry, "Full-spectrum interrogation of fiber Bragg gratings at 100 kHz

for detection of impact loading," Measurement Science and Technology, vol.

21, p. 094009, 2010.

[20] S. H. Yun, D. J. Richardson and B. Y. Kim, "Interrogation of fiber grating sensor

arrays with a wavelength-swept fiber laser," Optics letters, vol. 23, pp. 843-

845, 1998.

[21] S. H. Yun, D. J. Richardson, D. O. Culverhouse and B. Y. Kim, "Wavelength-

swept fiber laser with frequency shifted feedback and resonantly swept intra-

cavity acoustooptic tunable filter," IEEE Journal of Selected Topics in

Quantum Electronics, vol. 3, pp. 1087-1096, 1997.

[22] Y. Nakazaki and S. Yamashita, "Fast and wide tuning range wavelength-swept

fiber laser based on dispersion tuning and its application to dynamic FBG

sensing," Optics express, vol. 17, pp. 8310-8318, 2009.

[23] D. Chen, C. Shu and S. He, "Multiple fiber Bragg grating interrogation based on

a spectrum-limited Fourier domain mode-locking fiber laser," Optics letters,

vol. 33, pp. 1395-1397, 2008.




136

[24] E. J. Jung, C.-S. Kim, M. Y. Jeong, M. K. Kim, M. Y. Jeon, W. Jung and Z. Chen,

"Characterization of FBG sensor interrogation based on a FDML wavelength

swept laser," Optics Express, vol. 16, pp. 16552-16560, 2008.

[25] H. D. Lee, G. H. Kim, T. J. Eom, M. Y. Jeong and C.-S. Kim, "Linearized

wavelength interrogation system of fiber Bragg grating strain sensor based

on wavelength-swept active mode locking fiber laser," Journal of Lightwave

Technology, vol. 33, pp. 2617-2622, 2015.

[26] I. Photonics, Spectrometer design guide

http://ibsen.com/technology/spectrometer-design-guide/.

[27] T. Bodendorfer, M. S. Muller, F. Hirth and A. W. Koch, "Comparison of different

peak detection algorithms with regards to spectrometric fiber Bragg grating

interrogation systems," in Optomechatronic Technologies, 2009. ISOT 2009.

International Symposium on, 2009.

[28] J. Jiang, T. Liu, K. Liu and Y. Zhang, "Investigation of peak wavelength detection

of fiber Bragg grating with sparse spectral data," Optical Engineering, vol.

51, pp. 34403-1, 2012.

[29] A. Ezbiri, S. E. Kanellopoulos and V. A. Handerek, "High resolution

instrumentation system for fibre-Bragg grating aerospace sensors," Optics

communications, vol. 150, pp. 43-48, 1998.




137

[30] J. M. Gong, J. M. K. MacAlpine, C. C. Chan, W. Jin, M. Zhang and Y. B. Liao,

"A novel wavelength detection technique for fiber Bragg grating sensors,"

IEEE Photonics Technology Letters, vol. 14, pp. 678-680, 2002.

[31] D. W. Marquardt, "An algorithm for least-squares estimation of nonlinear

parameters," Journal of the society for Industrial and Applied Mathematics,

vol. 11, pp. 431-441, 1963.

[32] J. J. Moré, "The Levenberg-Marquardt algorithm: implementation and theory," in

Numerical analysis, Springer, 1978, pp. 105-116.

[33] T. E. Oliphant, "Python for scientific computing," Computing in Science \&

Engineering, vol. 9, 2007.

[34] J. J. Moré, D. C. Sorensen, K. E. Hillstrom and B. S. Garbow, "The MINPACK

project," Sources and Development of Mathematical Software, pp. 88-111,

1984.

[35] R. A. Caruana, R. B. Searle, T. Heller and S. I. Shupack, "Fast algorithm for the

resolution of spectra," Analytical chemistry, vol. 58, pp. 1162-1167, 1986.

[36] H. Guo, "A simple algorithm for fitting a gaussian function [DSP tips and tricks],"

IEEE Signal Processing Magazine, vol. 28, pp. 134-137, 2011.




138

Chapter 6 Conclusions and Future Work

139

Chapter 6

Conclusions and Future Work

This thesis studies the applications of FBG sensors for force and shape estimation for

flexible medical instruments.

A method to estimate the forces at the tip of a hollow needle has been presented in

chapter 3. It has been shown that this can be used to estimate the transition of needle

through tissue layers of varying stiffness. The analysis shows that gradients of the strains

measured by the sensor close to the tip can be used to determine the transition without

the need to depend on the absolute values of the strains. In future, the method can be

extended to also determine the compression of the tissue during needle insertion.

Need for shape estimation of flexible instruments and methods to do it have been

explored in chapter 4. Curvature estimation using FBG sensors bonded to a nitinol wire

have been found to significantly reduce the flexibility of the assembly. A new alternative

based on FBG sensors embedded in a PDMS filled tube has been presented. The

tolerance of FBG sensors to the curing process has enabled their use for the fabrication

of the shape sensing tube. However, the setup leads to a non-determinism of rotation

angle. It has been shown that an Artificial Neural Network based calibration can be used

to estimate the radius of curvature and hence aid in the reconstruction of shape for in-

plane bending. In future, the method can be extended to include estimation for out-of

plane bending as well and hence can enable complete 3D shape reconstruction.

Cost effective interrogation systems are as important as sensor characteristics for

widespread commercial use of FBG sensors in biomedical applications. This thesis has

Chapter 6 Conclusions and Future Work

140

shown that interrogation systems based on InGaAs linear detector arrays in combination

with computationally efficient curve fitting techniques can achieve good accuracies to

enable their usage for biomedical applications. In future, the system can be extended

with an approach to enable dynamic trade-offs of wavelength range versus resolution.

The benefits of small size and multiplexing of FBG sensors can enable the creation of a

unified setup as shown in figure 6.1 where shape sensing can be used in conjunction

with force sensing for aiding surgeons in complex procedures like Natural Orifice

Transluminal Endoscopy.

Fig 6.1 Proposed future device with integrated shape sensing tube and force sensing

needle

Flexible scope with shape sensing

Working Channel Needle with FBG

sensors

141

Appendix A

FBG sensor fabrication setup

Figure A.1: Setup used for fabricating FBG sensors

The beam from KrF excimer laser (Braggstar industrial) is focused on the fiber using a

cylindrical lens. The phase mask is aligned perpendicular to the laser beam. The fiber is

mounted on a translation stage with a motion controller of resolution 0.1µm and speed

of 20mm/sec. The phase mask is mounted on a stage which has x, y and z translation,

tilt and rotation adjustments. This enables very precise alignment. The alignment is

Phase Mask

KrF Laser

Linear Motion Stage

Motion Controller Camera

142

monitored using a camera mounted on top of the setup. During fabrication, the fiber is

placed in very close proximity to the phase mask. The entire setup is placed on a

vibration isolation optical table.

143

List of Publications

Kumar Saurabh, Venkoba Shrikanth, Bharadwaj Amrutur, Sundarrajan Asokan, M.S.

Bobji, “Detecting stages of needle penetration into tissues through force estimation

at needle tip using fiber bragg grating sensors”, J. Biomed. Opt., 21(12), 2016

Kumar Saurabh, V Shrikanth, Bharadwaj Amrutur, Sundarrajan Asokan, M.S. Bobji,

“Estimating needle-tissue interaction forces for hollow needles using fiber Bragg

grating sensors”, Proc. of. SPIE 9702, Mar 2016

Kumar Saurabh, Shanthanu Chakravarthy, G. K. Ananthasuresh, Bharadwaj Amrutur,

Asokan Sundarrajan, “Shape Estimation for Flexible Medical Instruments - An

Approach Based on FBG Sensors Embedded in a Biocompatible Polymer Filled

Tube”, IEEE/RSJ IROS Workshop on Navigation and Actuation of Flexible Instruments

for Medical Applications (NAFIMA), Hamburg, Oct 2015

Ananya, Shanthanu Chakravarthy, Kumar Saurabh and G. K. Ananthasuresh, “Shape

Estimation and Prediction of Locations of the Force on a Flexible Tube using

Strains at a Few Points”, 2nd International and 17th National Conference on

Machines and Mechanisms, 2015

Kumar, Saurabh, Bharadwaj Amrutur, and Sundarrajan Asokan. "Evaluation of fiber

Bragg grating sensor interrogation using InGaAs linear detector arrays and

Gaussian approximation on embedded hardware" Review of Scientific Instruments,

89.2 (2018)

An experiment is a question which science poses to Nature, and a

measurement is the recording of Nature's answer.

- Max Planck

force and shape estimation using fiber bragg grating ... · i would like to thank dr. abhijit lele...

Documents