MRI

ROBO

TIC

S INNOMOTION forPercutaneous Image-GuidedInterventions

Magnetic resonance (MR) guided percutaneousinterventions, such as biopsies, drainage, andinsertion of energetic probes for tumor ablation,have been developed and clinically demonstrated

with open-bore, low-field MR systems [1], [2]. MRI-guidedtreatment of spine diseases was achieved with a 0.2-T openmagnet [3]. In comparison, closed-bore, high-field MR imag-ing (MRI) scanners � 1 T have better spatial and temporalresolution, but patient access is more limited, and, hence, theyare less feasible for interventions and demanding for robotics.

Various image-compatible robotics have been developed,such as fluoroscopy and computed tomography (CT) image-guided kidney biopsies [4]. Chinzei et al. have introduced arobotic assistance system dedicated for the General ElectricSigna SP double doughnut open MRI [5], and Gassert et al.report on MRI-compatible robotics for interaction with humanarm motions [6]. Hitherto, no dedicated telemanipulator forhigh-field MRI- or CT-guided intervention has been published.Our system emanated from the lack of robotic instrumentationdesigned for the environment and unique ergonomics of ahigh-field close-bore magnet. The development of a fully MR-compatible robotic system started in 1998 at the Forschungs-zentrum Karlsruhe (FZK) in collaboration with the Universityof Applied Sciences Gelsenkirchen, and, in 2001, the GermanCancer Research Center also joined. The final product devel-opment was performed by the start-up company Innomedic.

The work shown in this report describes the development of arobotic assistance system that provides precise and reproducibleinstrument positioning inside the MRI or CT gantry on the basisof processed imagery. For the solutions developed, neither theMRI nor the CT systems were altered to any extent [7]. Percuta-neous interventions such as MRI-guided insertion of cannulaeand probes for biopsy, drainage, drug delivery, and energetictumor destruction were primary drivers in the robotic design.Applications involving the central nervous system have beenexcluded because of the demanding medical device regulationsand approval process. The developed system received the Con-formite Europeenne (CE) mark in 2005 and is marketed underthe brand name INNOMOTION. It is currently in clinical use forMRI-guided sciatic pain and facet joint treatments, biopsies,

drainages, and CT-guided osteosynthesis. This article reviewsthe development and evaluation of the system.

History of Development

First Generation: MIRA Manipulatorfor Interventional RadiologyThe initial design concept was derived from a telemanipulatorproject for endoscopic surgery [8]. Exploration of standardrobotics and commercially available robotic platforms andsurgical manipulators revealed the need for new MRI-compatiblecomponents. The demands of full electromagnetic compatibil-ity led to the use of new plastic materials, ceramics, and thedevelopment of new sensors and drives. Preliminary studies onthe process of interventional MRI and evaluation of workspace(movements, forces, required target precision, etc.) identifiedthe need for a nonconventional kinematic design that is spe-cific to close-bore MRI scanners and CT units.

The kinematics consists of a main robotic arm with 6 degreesof freedom (DoF) [7] that is attached to a stable orbital ring of580/490 mm (outer/inner) diameter, which is made of glassfiber-reinforced polyester. The orbit ring is mounted to the MRIpatient bed with specially made clamps that are designed to fitmost MRI and CT platforms. Studies with volunteers and involv-ing a design team have provided a better understanding of thesystem’s environment. Working process, work space dimensionsand geometry, measures of links, and range of articulation wereevaluated and approved with a 1:1 mock-up model (Figure 1).

A detailed list of specifications was established by means ofanalyzing the conventional MRI- and CT-guided proceduresin terms of the clinical process, devices used for handling andplacement of the patient, use of imaging sequences, communi-cation protocols for exchange of images, coordinates systems,coordinates transfer, supplying lines, and safety regulations.

The first prototype was driven by piezoelectric motors andwas designed to provide continuous positioning of the mainarm at the orbit ring through a friction-based drive. However,this version was discarded because of technological problemswith the drives. Piezoelectric drives induced signal noise dur-ing the MRI scanning process, and the electric power lines cre-ated the risk of inductive heating. In addition, backlash due togear play led to kinematic inconsistencies (Figure 2).

BY ANDREAS MELZER,BERND GUTMANN, THOMAS REMMELE,RENATE WOLF, ANDREAS LUKOSCHECK,MICHAEL BOCK, HUBERT BARDENHEUER,AND HARALD FISCHER

©PROJECT NEUROARM,UNIVERSITY OF CALGARY,CALGARY, ALBERTA

Digital Object Identifier 10.1109/EMB.2007.910274

Principles and Evaluation of this MR- andCT-Compatible Robotic System

66 IEEE ENGINEERING IN MEDICINE AND BIOLOGY MAGAZINE 0739-5175/08/$25.00©2008IEEE MAY/JUNE 2008

Second Generation: MRI- and CT-CompatibleAssistance SystemThe current kinematics (Figure 3) consists of an arm that is pneu-matically driven in 5 DoF [4]. The robot arm is attached to a 180�-orbiting ring that is mounted to the patient table of the scanner, andcan be manually prepositioned into the orbit region, at the angles0�,�35�,�67�, on either side of the orbit ring, depending on theregion of interest (e.g., spine, liver, kidney, breast). The arm isfixed with a spring-loaded bolt and secured with a screw.

Sensor-Actuator SystemThis kinematic design allows to be driven only by MRI-compatible pneumatic linear cylinders. The desired designcharacteristic for the pneumatic cylinder, to function over itsentire piston range with a particular force that was dependent onthe pressure, was solved with newly developed slow-motion con-trol of<0.1 mm/s, 146 mm range, and 295 N at 6 bar. The sametype of cylinder drives all five axes (Figure 4). Conventionalpneumatic drives are nonlinear and difficult to control. A specialdesign of the cylinders resulted in high dynamic friction (about40 N), avoiding collapse of the system from sudden pressure losscaused by leakage or other hardware failure. The design alsoresults in a static friction that is less than the dynamic friction,thus avoiding slip-stick effects of the actuators.

Active positioning measurements can be achieved via fiber-optic limit switches, rotational encoder (0.0088� resolution),and linear encoder (2-lm resolution) (Figures 5 and 6).

The actuators are controlled by an industrial controllerboard with a Super Harvard Architecture Single-ChipComputer Digital Signal Processor in the control PC and forexternal data acquisition in the system cart. An optical net-work (Info@SynqNet.org) transfers the data between both. A

standard proportional-integral-derivative (PID) controller isused for movement of the axes.

The control algorithm consists of an internal position loopand an outer velocity loop. To adapt the control algorithm tothe needs of a pneumatic actuator, the actuating variable isasymmetrically split to control two digital-analog convert-ers, one for each servo valve. The standard PID controlproves to be sufficient to handle potential nonlinearities ofthe pneumatic axes.

Application ModuleThe application module for guidance of coaxial probes (e.g., can-nulae for biopsies, radio frequency or laser probes, endoscopes,

Fig. 1. Design model of the first-generation MR-compatiblerobotic telemanipulator system (courtesy Forschungszen-trum Karlsruhe (FZK), Germany, 2000).

Fig. 2. The prototype MIRA driven by piezoelectric motorwas built at FZK, Germany.

Fig. 3. Schematic view of the INNOMOTION assistance sys-tem with five pneumatically driven DoF and two manualadjustments for prepositioning at the orbit (red arrow) andat the patient bed (green arrow).

Fig. 4. MR-safe pneumatic cylinders without metallic compo-nents in various lengths.

Fig. 5. MR-compatible rotational encoder has been developedby using polymers, ceramic materials, and optic components(courtesy Innomedic).

IEEE ENGINEERING IN MEDICINE AND BIOLOGY MAGAZINE MAY/JUNE 2008 67

etc.) provides 2 DoF. This design assures stable positioning ofthe instrument within a tool center point that is keeping the invar-iant point of insertion through the skin (Figure 7).

In conjunction with the two axes for movements in the toolcenter point (�40� rotation in transversal and, typically,�23�/þ70� in sagittal direction depending on the inclinationof the preceding axes), the instrument trajectory can be changedto other targets without moving the robot arm or repositioningthe arm at the orbit. A pneumatic drive has been developed toinsert the cannula in incremental steps of 1–20 mm, but itremained in experimental stage for future systems application[Figure 8(a) and (b)].

The application module (Figure 7) for clinical use providesmanual translation and rotation of the cannula. Because ofsafety reasons, any new position of the insertion trajectory or anew insertion point requires new planning of the puncture.

Human-Machine InterfaceThe concept of telemanipulation with haptic feedback wasexamined with an experimental setup using conventionalservo motors and force feedback. Trained and untrained vol-unteers were using a master control device for puncture ofvarious materials and tissues with conventional 18-G cannulae.

The master control device (MDC) trans-mitted the movements through a belt toan incremental sensor and to an electricmotor-driven cannula [9]. The hapticfeedback system must not be MRI-compatible because it is planned to beused outside the MR suite (Figure 9).

The master–slave concept for insertingcannulae and probes under image guidanceincluded the use of the pneumatic drive ofthe instrument. Because of regulatoryissues, liability problems, and high cost,this approach is currently not implementedin the first current product, but it will bedeveloped further for MRI-guided surgicalapplications and implemented if required.

The current human–machine interface comprises a flatscreen. Graphical user interface (GUI) for planning of targetand insertion point on a screen via mouse pad. Initial use of atouch screen was discarded because of problems in precisehandling of the graphic components on the screen (Figure 10).

MRI and CT images are sent via Digital Imaging and Com-munication in Medicine (DICOM) protocol to the INNOMO-TION PC, and planning of insertion and target point can beperformed throughout the imaging for double angulations ofthe insertion trajectory. The coordinates are sent to the system,and it moves the application module to the insertion point atthe patient.

Functional System SafetyThe system consists of highly redundant safety features accord-ing to the European Union standards EN 61508/SIL2. The mainfeatures are as listed:� external watchdog with time slice� pressure and voltage monitoring� internal watchdogs for input/output (I/O data) and con-

troller boards� monitoring of axis state by the controller board� monitoring of encoder state by the controller board� monitoring of signal quality of the optical encoders by

microcontrollers.Furthermore, different software measures are used to

guarantee the integrity of internal data storage and datacommunication.

In case of an error, the system goes into a safe state, i.e., theair supply is interrupted by closing the valves, and each axis isstopped by the emergency stop feature of the controller board.The friction keeps the system stable. During imaging, the can-nula-guiding sleeve can be mechanically disconnected to avoidinadvertent displacement of the probe.

Decontamination and SterilityThe system allows swab disinfection according to EuropeanUnion standards. For the procedures, a specially made sterilepolyethylene cover is placed over the front end. Subsequent tothe final images, the large ring surface coil is placed on the areaof insertion point and fixed with a prefabricated sterile drape witha central opening of approximately 6 cm in diameter.

The arm of INNOMOTION can be lifted up to ease cover-ing the arm of the system with the sterile polyethylene cover,and square opening at the front provides insertion of a steri-lized reusable polyether ether ketone (PEEK) bracket that

Lin. Sensors Rot. Sensors Actuators

Fig. 6. Schematic view of the arrangement of MR-compatible optical sensors andpneumatic actuators.

Fig. 7. The kinematics of the AMO provides a stable positionof the insertion point (red). Spherical marker gadoliniumcontrast agent (blue) filled for localization on MRI (courtesyInnomedic).

IEEE ENGINEERING IN MEDICINE AND BIOLOGY MAGAZINE MAY/JUNE 200868

holds the sterile sleeve that is premounted to the cannula (seeFigure 11). The free area of skin is disinfected twice withstandard disinfectant. The arm is pneu-matically driven back to the insertionpoint, sleeve and cannula are attached,and the cannula then inserted throughthe skin.

The sterile cover and the sleeves forguiding the cannulae are disposableones. Both bracket and sleeve holderhave to be cleaned, sealed in an appro-priate pouch, and sterilized after use.

Interventional ProcedureThe patient can be placed in the pre-determined position suitable for theintervention (supine, prone, or lat-eral). According to the preinterven-tional images and according to theanatomic regions of interest, the tableand INNOMOTION are positionedaccording to the projection of thelaser light of the MRI. The system isprepositioned and firmly attached tothe table with clamps. The laser light isswitched off, and the table is movedinto the MRI bore until the position ofthe laser line matched with 0 of the z axis of the MRI. Planningof interventions is performed by using fast gradient-echosequences in transverse, sagittal, or coronal orientation; e.g.,TrueFISP or fast field echo. Suitable slices are selected and sentvia the network in DICOM format to the computer of the assis-tance system. Insertion site and a target point are selected on theGUI monitor, and the coordinates are sent to the control unit.The arm moves into a close position to the predetermined inser-tion site (15-mm distance). Because the INNOMOTION sys-tem is always fixed parallel to the MR table by rails, the onlytransformation between the coordinate systems of the MR andINNOMOTION is a linear translation but no rotation. Thus, thedescribed technique for registration of INNOMOTION in theimage space is based on MRI of the contrast-filled marker at theapplication module (Figure 7). The INNOMOTION system isprepositioned by coarsely aligning the application module withthe laser light of the MRI system. Then, the system is refer-enced with the coordinate system of MR scanner using fourspherical markers at the application module (AMO) and a spe-cific sequence three-dimensional gradient echo [flip angle(FA): 50, time of repetition (TR): 13, time of echo (TE): 6] intransverse 40-mm and coronal 20-mm orientation. The positionof the markers in image space is evaluated by the system’s soft-ware by means of image processing algorithms [median filter-ing, segmentation by adaptive thresholds, calculation of objectfeatures of the segmented markers (i.e., perimeter and area)].The segmented object features are compared with the expectedfeatures to validate the quality and reliability of the markers.This position is transferred into the INNOMOTION coordinatesystem and compared with the expected position of the markerscalculated by a forward transform using the known joint posi-tions of the axis. Any residual deviation less than 1 mm isaccepted and used for an offset compensation. After markerdetection, the application module moves the tool center point tothe insertion site on the skin and adjusts the insertion trajectory

to the planning. The cannula can be inserted through a guidingsleeve or an open angle (Figure 11).

Fig. 9. Master-slave control device for remote-handled can-nula insertion with force feedback for tissue discrimination[9] (courtesy Forschungszentrum Karlsruhe [9]).

Fig. 10. The GUI (first generation) provides planning of inser-tion on the MR images sent via DICOM.

(a) (b)

–3,0

–2,0

–1,0

0,0

1,0

2,0

3,0

–3,0 2,0 –1,0 0,0 1,0 2,0 3,0

x (mm)

y (m

m)

Fig. 8. (a) Application module with the prototype of a pneumatic gripping and advanc-ing mechanism. (b) An experimental trial with the insertion of a 19-G stainless steel can-nula through fresh porcine abdominal wall revealed a target precision of �0.53 mm(courtesy Forschungszentrum Karlsruhe 2002).

IEEE ENGINEERING IN MEDICINE AND BIOLOGY MAGAZINE MAY/JUNE 2008 69

Interventional InstrumentsA variety of cannulae and other instruments of differentlengths and sizes have been developed [10] and are now avail-able with premounted guiding sleeve.

The application module is suitable for the majority of MRI-compatible coaxial instruments and probes by using custom-made sleeves provided by Innomedic.

Experimental EvaluationMRI compatibility has been achieved through testing ofall components and the complete operating system in

different field strengths: 1.5-T MR scanner (MagnetomSymphony and Espree) and 1.0-T Gyroscan and 1.5-TIntera [11] (Figure 12).

Mechanical targeting precision has been determinedaccording to DIN EN ISO 9282:1998 with a stereo optic sys-tem over 100-mm target distance (mean deviation < 0.5 mm/maximum deviation < �1.5 mm for each coordinates). Bycontrast to any experimental device, INNOMOTIONachieves this precision in serial production according toISO 9000. The histograms [Figure 13(a) and (b)] show an ex-ample for a calibration measurement.

The deviation is defined as the difference between the nom-inal position (i.e., the commanded position in space withrespect to the basis coordinate system of INNOMOTION) andthe actual position (i.e., the measured position in space). Thetool tip point is defined as the tip of a virtual instrument at100-mm distance from the tool center point.

The process chain MRI adds approximately �0.5–1 mmsystematic error because of pixel size, B-field inhomogeneity,and gradient deviation depending on the MRI system used.

The MRI procedures were performed with a 1.5-T SiemensMagnetom Symphony, 1.0-T Gyroscan, and 1.5-T Intera onex vivo organ models with fresh porcine kidney embedded ingelatin (Figure 14).

The evaluation of target precision has been proven duringMRI-guided percutaneous interventions in a porcine modelapproved by the institutional review board of the University of

Heidelberg. The animals (four 3-month-old domestic pigs 30–40 kg) wereplaced under general anesthesia (iso-flurane) in prone position on the patienttable, and a surface coil was fixedaround the planned insertion site lateralto the spine. On T1- and T2-weightedplanning images, the appropriate regionof trajectory was defined on the GUI ofthe INNOMOTION control computer,and the robot arm then moved andoriented the needle holder to the inser-tion point accordingly. MR-compatibletitanium Grade 4 cannulae (20 and22 G) were then manually inserted. Sub-sequent to an initial insertion of ap-proximately 10 mm, the table wasrepositioned into the MRI bore and con-trol images were acquired. The interven-tion was completed within the magnetfrom the rear opening, where an MR-compatible in-room monitor wasplaced. During insertion of the needle,real-time MR images were acquired to

Patient

System Trolley

Pneumatic SupplyMRI Console and Graphic User Interface (GUI)

System at MRI Patient Bed

Fig. 12. Setup for MRI-guided procedures as it has been evaluated for Siemens andPhilips MRI platforms.

Fig. 11. Varieties of cannulae and a biopsy gun withassembled sleeve and guiding elements for clinical use.

The application module for guidance of coaxial

probes provides 2 DoF and assures stable

positioning of the instrument within a

tool center point.

IEEE ENGINEERING IN MEDICINE AND BIOLOGY MAGAZINE MAY/JUNE 200870

control the insertion. To visualize advancement of the cannulain the tissue, fast gradient-echoes sequences (TR ¼ 4.4 ms,TE ¼ 2.2 ms, FA ¼ 70�, TA¼ 0.7 s) were used. At the desiredregion of interest (nerve root, celiac plexus), spin echo imageswere acquired for verification of the cannula position through atest bolus of Gadolinium-based contrast agent solution (1 mmol/l sterile physiological saline solution). The injection was done

Intervention0,5 mm, 0,5°

Fig. 15. Overlay image on the INNOMOTION screen for eval-uation of target precision of 20-G cannula insertion at theporcine celiac plexus (GUI first generation).

Fig. 14. Target-precision of MRI-guided insertion of 20-G tita-nium cannula (Daum) into a porcine kidney embedded ingelatin.

(a)

(b)

4

3

2

Fre

quen

cy

1

0–1 –0.5 0

Deviation x (mm)0.5 1

3

2

Fre

quen

cy

1

0–1 –0.2 0

Deviation z (mm)

Deviation Nominal/ActualPosition (Tool Center Point)

0.2 0.4

3

3

1

0–1 –0.5 0

Deviation y (mm)0.5 1

Deviation Nominal/ActualPosition (Tool Tip Point)

3

2

Fre

quen

cy

1

0–1 –0.2 –1

Deviation z (mm)0.2 0.4

4

3

Fre

quen

cy

Fre

quen

cy

1

2

0–1 –0.5 0

Deviation x (mm)0.5 1

4

3

2

1

00 –0.5

Deviation y (mm)1 1.5

Fig. 13. (a) Mechanical target precision of tool center pointat the skin insertion site. (b) Target precision of the tool tippoint, which is defined as the tip of a virtual instrument at100-mm distance from the tool center point.

Fig. 16. Bilateral facet joint treatment with 40 mg triamcino-lone and 10 ml mepivacain. Overlay of the planning imagereveals target precision [20-G Daum titanium cannula (GUIsecond generation)].

INNOMOTION has received CE mark and is

currently in clinical use for MRI-guided sciatic

pain and facet joint treatments, biopsies,

drainage, tumor ablation or embolization, and

CT-guided osteosynthesis.

IEEE ENGINEERING IN MEDICINE AND BIOLOGY MAGAZINE MAY/JUNE 2008 71

under real-time MRI (TR¼ 1.8 ms, TE¼ 4.3 ms, TA¼ 0.5–0.8 s,FA¼ 20�) to visualize the drug distribution.

Final therapeutic injection of 10–25 ml with contrast-dottedMepivacainhydrochlorid (Scandicain, 1%) was performed.Precision of insertion point and insertion angle has been deter-mined by using overlays of the preinterventional images withthe new MRI image (Figure 15).

All procedures were then completed successfully; e.g.,injection at sympathetic chain, sciatic nerve, and coeliacplexus. The direct MRI control with fast sequences techniquesallows correction of the insertion path in case of deteriorationdue to anatomic structures. The insertion site and the insertionangle have been evaluated by manual measurement on over-lays of the planning image of INNOMOTION and the subse-quent MR control image (Figures 15 and 16). Position andorientation of all cannula insertions were appropriately visual-ized on axial MRI images. Precision of insertion site in axialplane was �1 mm (minimum of 0.5 mm and maximum of3 mm). Angular deviation in the transverse plane of the cannu-lae shows�1� with minimum of 0.5� and maximum of 3�.

On the basis of these results the system received a CE markfor image-guided percutaneous interventions, except for thecentral nervous system, in 2005.

First Clinical TrialThe MRI procedures were performed with a 1.5-T Intera. Clini-cal evaluation of target precision and safety has been conductedduring MRI-guided percutaneous interventions on 16 patients(four females and 12 males) with informed consent.

All patients had previous MRI scans of the spine and havebeen treated via CT guidance at the same segment. MR-compatible titanium Grade 4 cannulae (20 and 22 G) werethen manually inserted to a depth of 4.8–6.7 cm. Gradient-echoes sequences (TR ¼ 4.4 ms, TE ¼ 2.2 ms, FA ¼ 70�,TA ¼ 0.7 s) were used for cannula guidance and drug instilla-tion of 5 ml mepivacain and 40 mg triamcinolone. Precision ofinsertion point and insertion angle has been determined byusing image overlays (Figure 16).

All interventions were successfully completed. Positionand orientation of all cannula insertions were appropriatelyvisualized on axial 3-mm MRI images. MR on table time was45 s to 11 min, and the last five procedures could be per-formed in less than 15 min. Apart from minor side effects ofincreased sweating in one patient after two procedures and

prolonged menstruation in one patient aftertwo procedures, no major adverse eventshave been noted.

Conclusion and OutlookCross-platform MRI compatibility of mecha-tronic systems for MRI-guided cannula in-sertion can be achieved by using polymers,ceramics, pneumatic drives, and optoelec-tronic sensors. Because the cannula is cur-rently to be advanced manually, the access isdifficult if insertion is done inside the mag-net. Thus, the direct control of the insertionunder real-time MRI is recommended as tobe able to correct insertion in case of deterio-ration of the cannula and to precisely posi-tion the tip of the cannula in the volume ofinterest. To ease the procedure, tip-tracking

techniques have been evaluated [12] (Figure 17).

AcknowledgmentsThe development of INNOMOTION by Innomedic GmbH,Herxheim, has been financially supported by the FZK, BASFInnovations Fonds, Ludwigshafen; tpg, Bonn; KfW, Frankfurt;and WFT, Mainz. The animal trial has been performed at andsupported by the Deutsches Krebsforschungzentrum DKFZ,Department of Medical Physics in Radiology, Heidelberg. Wegratefully acknowledge the outstanding work and commitmentof the following teams: Innomedic, Herxheim: Andreas Berger,Stefanie Gutmann, Lars Wischnewski, Horst Steigner;Research Center Karlsruhe: Sandra Boscan, Heinz Becker,Helmut Breitwieser, Lothar Gumb, Hartmut Gutzeit, ThomasHoehn, Heinz Junker, Marco Klein, Sven Koehn, HolgerKrause, Martin Mark, Georg Prokott, Marco Sidor, Udo Voges,Oliver Wendt; Deutsche Krebsforschungszentrum DKFZ:Hamid Ghaderi (University of Heidelberg), Hendrik Zimmer-man, Wolfhard Semmler; FH Gelsenkirchen: Peter Bremer,Thomas Bertsch, Waldemar Zylka; Marien Hospital Buer, Ger-many: Gert Lorenz, Wolfram Triebe, Sabine Matthay.

Andreas Melzer qualified in dentistry in1989 and received his M.D. degree in 1993.He has 20 years’ experience in the develop-ment of medical technology for laparoendo-scopic surgery, interventional radiology,interventional and intraoperative MRI,mage-guided robotics, surgical instrumen-tation, surgical robotics, and Nitinol devi-

ces. He is the coeditor of three medical journals, the cofounderand partner of six startup companies in the medical-technologybusiness, and a consultant for major vendors in medicine. Hehas organized and chaired various medical conferences and isthe board member of six medical and technical societies. Cur-rently, he is a professor at the Department of Medical Technol-ogy and is the director of the Institute for Medical Science andTechnology (IMSaT), a joint venture of the Universities ofDundee and St. Andrews, Scotland.

Bernd Gutmann studied physics at the University of Karls-ruhe. He received a diploma degree for his work on opticalmetrology where he combined in-plane pattern correlationwith out-of-plane fringe projection techniques. He continued

Fig. 17. Work in progress at DKFZ, Heidelberg. MRI tip-tracking techniquesprovide automatic slide orientation of the MRI according to the applicationmodule [12].

IEEE ENGINEERING IN MEDICINE AND BIOLOGY MAGAZINE MAY/JUNE 200872

his research at the Institute for MechanicalEngineering, where he received his Dr.-Ing.for his research in phase unwrapping ofhighly distorted interference patterns vianonlinear optimization techniques. As theproject manager, he is responsible for thedevelopment and production of INNOMO-TION at the Innomedic GmbH.

Thomas Remmele received diplomas inmechanical engineering and computer engi-neering in 1989 and 1992, respectively.Since 1993 he has been working on medicalrobotics in the Research Center Karlsruhe.He is a named inventor on more than 15patents since he founded the company Inno-medic GmbH in 1999. Innomedic devel-

oped in cooperation with the Research Center Karlsruhe andAndreas Melzer the fully MR-compatible robotic systemINNOMOTION. He is coauthor of more than 30 publicationson medical robotics. He has given several invited lectures atmedical and bioengineering conferences.

Renate Wolf is a specialist nurse in anesthe-siology and intensive care, and she has re-ceived degrees in sales and marketing. After16 years of experience in medical industry,as sales and marketing manager at Johnson &Johnson and St. Jude, she joined InnomedicGmbH and was responsible for marketing,sales, installation, and first clinical applica-

tions of INNOMOTION in Gelsenkirchen, Frankfurt, Leipzig,Basel, Essen, Vienna, and Innsbruck from 2003 to 2007.

Andreas Lukoscheck studied biomedicalengineering at the University of AppliedSciences in Gelsenkirchen and worked forComputer Motion, Inc., California. Hereceived his diploma degree for his workon process and economical evaluation ofCT and MRI image-guided interventionswith robotic systems at the University of

Applied Sciences in Gelsenkirchen. He has been the productmanager of INNOMOTION for four years. His special focusis on the workflow, usability, and reimbursement of INNO-MOTION. He is currently working for CT marketing at Tosh-iba Medical Systems, Germany.

Michael Bock studied physics at the Uni-versities of Braunschweig and Heidelberg,and he received his diploma degree for hiswork on electron impact excitation at theMax-Planck-Institute for Nuclear Physics inHeidelberg. For his Ph.D. thesis in physics,he joined the German Cancer Research Cen-ter (DKFZ) where he studied MRI flow-

measurement techniques. He is currently the head of a researchgroup for Interventional MRI at the DKFZ where he works onvarious aspects of MRI pulse sequences and hardware.

Hubert Bardenheuer received his medical degree at theRheinisch-Westfalische Technische Hochschule in Aachen,

Germany in 1980, then joined the Depart-ment of Physiology at the University ofMunich. From 1983 to 1984, he was researchassociate at the Department of Physiology,Michigan State University. He later becamea research assistant at the Clinic of Anesthe-siology at the University of Munich, special-izing in anesthesiology and intensive care

medicine. He was awarded a habilitation in the field of anesthesi-ology in 1991. In 1993, he became professor of anesthesiology atthe University of Heidelberg. Currently, he is working in theclinical and scientific field of acute and chronic pain therapy aswell as in palliative care medicine.

Harald Fischer studied electronics andbiomedical engineering at the Universityof Karlsruhe, and he received his diplomadegree for his work on haptic graphicalfeedback for minimally invasive surgery atthe FZK. For his Ph.D. thesis in medicalengineering, he worked on perception andtactile feedback for minimally invasive

surgery at the FZK. He is currently working on biologicalinterfaces and is the chief executive officer of a start-upcompany out of the FZK. Additionally, he is auditor for qual-ity management systems for medical devices.

Address for Correspondence: Andreas Melzer, Institute forMedical Science and Technology (IMSaT), University of Dundee,Dundee DD21FH, Scotland. E-mail: a.melzer@dundee.ac.uk.

References[1] J. S. Lewin, J. L. Duerk, V. R. Jain, C. A. Petersilge, C. P. Chao, andJ. R. Haaga, ‘‘Needle localization in MR-guided biopsy and aspiration: effects offield strength, sequence design, and magnetic field orientation,’’ AJR Am. J.Roentgenol., vol. 166, no. 6, pp. 1337–1345, 1996.[2] J. S. Lewin, S. G. Nour, and J. L. Duerk, ‘‘Magnetic resonance image-guidedbiopsy and aspiration,’’ Magn. Reson. Imaging, vol. 11, no. 3, pp. 173–183, 2000.[3] A. Melzer and R. Seibel, ‘‘MR guided therapy of spinal diseases,’’ Minim.Invasive Ther. Allied Technol., vol. 3, no. 2, pp. 89–93, 1999.[4] K. Cleary, A. Melzer, V. Watson, K. Kronreif, and D. Stoianovici, ‘‘Interven-tional robotic systems: Applications and technology state-of-the-art,’’ Minim.Invasive Ther. Allied Technol., vol. 15, no. 2, pp. 101–113, 2006.[5] K. Chinzei, N. Hata, F. Jolesz, and R. Kikinis, ‘‘MRI compatible surgicalassist robot: System, integration and preliminary feasibility study,’’ in Proc. MIC-CAI, 2000.[6] R. Gassert, R. Moser, E. Burdet, and H. Bleuler, ‘‘MRI/fMRI compatiblerobotic system with force feedback for interaction with human motion,’’ IEEE-ASME Trans. Mechatron., vol. 11, no. 2, pp. 216–224, 2006.[7] B. Gutmann, L. Gumb, M. Goetz, U. Voges, H. Fischer, and A. Melzer,‘‘Principles of MR/CT compatible robotics for image guided procedures,’’Radiol. Suppl., vol. 225, 0032CEVI, p. 677, 2002.[8] A. Melzer, M. O. Schurr, W. Kunert, G. Buess, U. Voges, and J. U. Meyer,‘‘Intelligent surgical instrument system ISIS: Concept and preliminary experimen-tal application of components and prototypes,’’ Endosc. Surg. Allied Technol.,vol. 1, no. 3, pp. 165–170, 1993.[9] H. Breitwieser, S. M. Boscan, H. Becker, U. Voges, H. Junker, andA. Melzer, ‘‘Feasibility of manual control device for robotic assisted needle inser-tion in CT and MRI,’’ Radiol. Suppl., vol. 225, p. 155 (192VI-P), 2002.[10] A. Melzer, A. M. Schmidt, K. Kipfmuller, M. Deli, D. Stockel, D. H.W. Gronemeyer, and R. M. M. Seibel, ‘‘Perequisites for magnetic resonanceimage-guided interventions and endoscopic surgery,’’ Minim. Invasive Ther.Allied Technol., vol. 5, no. 3, pp. 255–262, 1996.[11] A. Melzer, B. Gutmann, A. Lukoschek, M. Mark, W. Zylka, and H. Fischer,‘‘Experimental evaluation of an MRI compatible telerobotic system for CT MRIguided interventions,’’ Radiol. Suppl., vol. 226, p. 409, 2003.[12] H. Zimmermann, S. Muller, B. Gutmann, H. Bardenheuer, A. Melzer,R. Umathum, W. Nitz, W. Semmler, and M. Bock, ‘‘Targeted-HASTE imagingwith automated device tracking for MR-guided needle interventions in close-boreMR systems,’’ Magn. Reson. Med., vol. 56, no. 4, pp. 481–488, 2006.[13] Sterilization of Medical Devices—Requirements for Medical Devices to BeDesignated ‘‘Sterile’’—Part 1: Requirements for Terminally Sterilized MedicalDevices, DIN EN 556-1.

IEEE ENGINEERING IN MEDICINE AND BIOLOGY MAGAZINE MAY/JUNE 2008 73