1

Biomedical circuits and systems torecuperate neuromuscular functions

Mohamad Sawan, P. Eng., Ph.D., FCAE,Canadian Research Chair on Smart Medical Devices

PolySTIM Neurotechnology LaboratoryDept. of Elec. Eng., École Polytechnique de Montréal

m.sawan@polymtl.ca

PART I

IEEE CASS DLPJanuary 2004

w Introductionn Purpose and motivation

w Microelectronics activitiesn Design, construction & test of smart medical devices (SMDs)n Typical architecture of an SMD

Focus on two main case studiesn Neurostimulation to recuperate the bladder functionsn Intracortical neurostimulation to create vision for the blind

Required biocircuits & biosystemsResearch team and collaborationsConclusions

2

Biomedical circuits and systems to recuperateneuromuscular functions

OUTLINE

M. Sawan3

w 1947, transistor allows circuit designs suitable for implants.

w 1952, first external pacemaker which was the size of a tableradio of the time and was powered by 110 VAC.

w 1957, electrical stimulation in the inner ear of the acousticnerve in a totally deaf human was reported.

w 1960, totally implanted pacemaker (Buffalo).

w 1961, peroneal nerve stimulator for foot drop in hemiplegics.w 1968, implantation of an electrodes array on the visual cortex.

w 1980, first microchip was used to design small pacemakers.

w 1984, FDA approved the first cochlear implant for adultsw 1991, recording of neural activities

w 1992+, FES is widely used in several applications.

Main technological breakthroughof FES based SMDs

M. Sawan

4

w Pacemaker: More than 350,000 pacemakers annually;

w Defibrillator: More than 50,000 implants annually;w Cochlear implant: More than 50,000 people worldwide;

w Functional stimulation: Recuperate hands and legsmovements of paralyzed patients;

w Stimulation for the treatment of Parkinson, respiration,pains, scleroses, etc;

w Automatic liberation of insuline in diabetics;

w Neuronal activities recording for better control;

w Devices for vision: Creation of adequate vision for blinds;w Other implantable microsystems.

Main technological breakthrough(Cont’d)

M. Sawan 5

Features:

Modulator

Demodulator

Receiver

Powersupply

Measurementand

digitalizationcircuitry

Stimuli generator

Current sources

Physio-logicalstimuli

Teststimuli

Clk

Pardata

MUX

DEMUX

Backtelemetry

Stimulation signals

Returned informations

- ETC impedance measurement- Charge per phase estimation- ETC voltage monitoring- Neural signal recording

Bi-directionalRF link

elec-trodes

Implant

Dataprocessing

Skin

Externalcontroller

Main

controller

Smart medical devices Proposed topology

M. Sawan6

Global view

Proposed SMD topology(cont’d)

C1

L1 L2 C2 C3

VoltageRegulator

Load ShiftKey (LSK)

DataDemodulator

ADCEncoder MUX

Controller/Stimulator

Charge Pump 1

Charge Pump 2

Charge Pump 3

Charge Pump 4

Vdd1

Vdd2

Vdd3

Vdd4

Vdd

Analogfrontend 1

Analogfrontend n

Electrodes

Start-upCircuit

ProtectionCircuit

Clock GeneratorCircuit

SwitchingRegulator

ASK Demodulator /DAC/Decoder

DataModulator

PowerAmplifier

Battery

Vdd

SignalProcessor

ClockGenerator

External Controller

Implant Circuit

Off-Chip On-Chip

M. Sawan

regulatorrectifierrflinktotal hhhh ⋅⋅=

7

Power Transfern Efficiency and safety.

Data transmissionn Manchester based codecn PSK based codec: BPSK

Demodulator (COSTASLoop)

Link features

RF LINK to Transfer Power & Data(cont’d)

Without Feedback: 12.03%With Feedback: 22.04%

• Full duplex mode• Transfer rate:

-LSK: 200Kb/s-BPSK: 1Mb/s

• PSK demodulator consumes only529 mW.•QPSK may double the data rate.

M. Sawan

VCO Low Pass Filter

Data Out

Data In

PhaseShifter

90

I branch

Q branch

Low Pass Filter

Low Pass Filter

)sin()( 1 1 qw +ttm

)sin(2 2 1 qw +t

)cos(2 2 1 qw +t)sin()( 2 1 qq -tm

)(2sin2)( 2 1 qq -tm

)cos()( 2 1 qq -tm

• Recover thecarrier and thedata in the sameloop

regulatorrectifierrflinktotal hhhh ⋅⋅=

The bladder controller: Dual-stimulator

• Lower urinary tract deficiency followinga spinal-cord injury;- Incontinence and/or retention.

• Multiples complications at severallevels (Sphincter, bladder and kidneys)due to repetitive catheterizations;

• Bladder hyperreflexia (hypertrophy or atrophy) hardly curable with traditional medicine;

• Electrical stimulation of the detrusor muscle allows partlyvoiding and may induce kidney complication.

8M. Sawan

• Conventional electrical neural stimulation (ENS)induces contraction on both detrusor and sphincter,thus preventing from complete voiding;

• Conventional ENS combined with nerve rhizothomy isnot accepted by patients;

• Early ENS tends to reduce the hyperreflexia phase,while limiting damage in the whole urinary system;

• Selective neural stimulation proposed by ourteam allows efficient voiding of the bladder;

• Permanent low frequency combined withselective stimulation are proposed by ourteam to improve bladder function.

9

The bladder controller: Dual-stimulator

M. Sawan

2

• Selective stimulation allows improvement in voiding;• This technique diplexes high and low frequency stimuli

to activate both categories of sacral nerve fibres:

- HF stimuli for somatic fibres which innervate the sphincter.

- LF stimuli for parasympathetic fibres which innervate thedetrusor.

10

The bladder controller: Dual-stimulatorSelective neural stimulation

M. Sawan

l First selective stimulator includes :‡ Implant of 3.5 cm of diameter built around field programmable

devices (FPD) and other discrete components;‡ Shape Memory Alloy (SMA) based cuff electrodes to facilitate

connecting to sacral roots;‡ User friendly external controller based on FPD and other

discrete components.

SMA Electrode CuffSelective Implantable Stimulator

11

The bladder controller: Dual-stimulatorSelective neural stimulation

M. Sawan

l Proposed dual-stimulator includes :-Permanent LF stimulation : Low amplitude train:

- Prevent the bladder hyperreflexia- Maintain the bladder shape- Selective stimulation.

l The system includes:- Battery to power thepermanent stimulator;- RF link for the selectivestimulation.

Amp

PW

1/Freq

Ton Toff

12

The bladder controller: Dual-stimulatorPermanent/Selective stimulations

M. Sawan

Lesion

Bladder

Bipolar cuff Electrode

Dual-stimulator Implant User Interface RF Emitter

Data

Param Stim #530Hz 175us

0.9mA

Skin

Power

External Controller

Decoder

DACCurrent

CommutatorCurrentSource

Controller#1

(FPGA)

Controller#2

(PIC)

Bus Controller

DSR

SYS_ON

DSR READY

READY

Z_CTRL

DIN/SCLK/CS

UP/DOWN

3

2

DIN/SCLK/CS

UP/DOWN

Switch#1

SYS_ON

SYS_ON

SYS_OFF

RF_PWR_EN

BAT_PWR_EN

BAT_PWR_EN

RF_PWR_EN

AmDemodulator

RF_PWR

DATA_IN

Switch#2

BAT_PWR

Switch#3

PIC_PWR

PIC_PWR

SYS_OFF

Nerve

DIN/SCLK/CS

UP/DOWN

SWT_PWR

SWT_PWR

Antenna

Battery

3

2

32G_CLOCKMANCH_IN

13

The bladder controller: Dual-stimulatorImplant description

M. Sawan

• External controller• Subcutaneous Implant (~ 2.5cm)• One cuff electrode pair

Bladder dual neurostimulation: Results

l Intravesical and intraurethral pressures measurements andElectromyographic (EMG) recording of the pelvic floor muscles wereperformed.

Voiding by selective stimulation

31%

88%

34%

69%

12%

66%

cathcath0 ml

100 ml

200 ml

50 ml

150 ml

250 ml

MeanBladderVolume

Residual

Voided

LF LFLF+HF+permanent

14M. Sawan 15

w 1960s, few groups began to investigate the possibilities ofexploiting the phenomenon of creating points of lights;

w 1990s, Researchers at NIH demonstrate that intra cortexstimulation allows to generate phosphenes;

w Progress in microelectronics and microfabrication motivateresearchers to explore several approaches;

w Dobelle institute, New York, early in 2000 presented a patientholding a PC which drives a percutaneous connector towardthe skull to extracortical visual region;

w We start this project in 1996n Early 2000, a prototype has been completed to prove the

feasibility of a visual cortical stimulatorn Miniaturized implant version is being achieved.

The visual cortical stimulator Evolution

M. Sawan

16

The visual cortical stimulator Approaches

Researchers worldwide are developing solutions Several approaches are being explored:n Artificial retina;n Cuff electrodes around the optic nerves;n Tongue stimulation;n Surface stimulation of the visual cortex;n Intracortical stimulation of the visual cortex.

Intracortical stimulation does not depend on the health ofthe eye or optic nerve; also, it requires stimuli at low currentlevel. It can be the predominant one:n Artificial retina requires functional retina and optic nerve;

n Surface stimulation of the visual cortex is imprecise;n Other techniques only allow sensory substitution .

M. Sawan 17

Artificial retina / Extracorticalstimulation

Experimentationn Stable, non flickering phosphenes;n Able to distinguish shapes;n Lower threshold in macular region;n Resolution restricted by electrode shapen Mechanical properties challenges.

Extracortical surface stimulationn Images feed from a digital camera to a belt-

mounted processor;n commands send through the skull, and into

the visual cortex

Optic nerve stimulationn Fixed, stable phosphenesn Many phosphenes with few electrodesn Phosphene characteristics widely variable.

Tongue stimulation, and auditory ortactile input.

M. Sawan 18

Intracortical stimulation

Intra-cortical stimulationn Better resolution

n Better control on phosphenes

n Lower power

n Lower chances of seizure.

Subject: 38 year old womann Parameters affect brightness, not position

n Elicits stable phosphenes

n Threshold ~15 µA (as low as 1.9 µA)

n Short term accommodation.

Experiments in Monkey at IITn Recently, report good news.

M. Sawan

3

19

The visual cortical stimulator The whole proposed system

StimulatorChips

InputImage

Camera

1

2

Image Processor & Power source

3

WirelessTransmitter

4

ControllerChip

6

7

5Antenna

Cortex

The system includes 4 main components:n Camera mounted onto a pair of glasses;n External controller to process the images and commands;n Bidirectional data & energy transmitted by RF link;n Implant to stimulate the region of the cortex controlling vision.

M. Sawan 20

The visual cortical stimulatorUndertaken research projects

Front-end(Camera& DSP)

Front-end(Camera& DSP)

RxRx InterfaceInterfaceTxTx

RF Data and PowerRF Data and Power

StimulatorStimulator

System level optimization:n Front-end for efficient image processing;n Tx / Rx for power efficiency and data rate;n Power consumption, size, safety, biocompatibility;n Interface for shape and surface.

System level optimization:n Front-end for efficient image processing;n Tx / Rx for power efficiency and data rate;n Power consumption, size, safety, biocompatibility;n Interface for shape and surface.

Timing HardwareTiming Hardware StimulatorStimulator InterfaceInterface

ExternalComponents

ExternalComponents

ImplantedComponents

ImplantedComponents

IsolationIsolation

1011000101

&&Development

board

Computer Matrixof

electrodes

M. Sawan

Prototype (VCS)

PVCS emulatinga 16x16 matrix

21

VDB source

Inverse Stim. Site to PVC Mapping

Display

Imageacquisi-

tion

Program-ming GUI

Imagefitting

External components

Implantablecomponents

Skin

VisuotopicMapImage

enhance-ment

InterfaceModule

Stim.Module

Stim.Module

Trans-mitter

Encoding& timing

PVC toSSA

mappingOrdering

Low levelPulse

sequencing

Visual FieldEmulator

Display

ArtificialVisuotopicMap gen.

Data collected fromexperimentation

SSA toPVC

mapping

...

......

Electrodes

Config.

Direct Mapping

Inputs

Low LevelData Manag. Implant

The visual cortical stimulator Integrated smart devices

•• Block diagram

M. Sawan

22

Components of Multi-modulesn Stimulator Modules (SMs)w Multi-channel stimulationw Electrode-tissue and

stimulator monitoring

n Interface Module (IM)w Bidirectional wireless

Communicationw Implant controlw Power recovery / management

w A/D Conversion of monitoring data

The visual cortical stimulatorImplant architecture

M. Sawan23

Implant architectureStimulator Module

Features (flexibility/safety)n Current Constant Stimulation

w Monopolar, Bipolarw Single/Multiple paths

n Configurable Comm.protocolw One to 25 bits per pulse.

Enables the use of a LF Clkw System power saving

n Voltage Monitoring

n RAM Configured Parametercontinuous error checkingw Disables stimulation in case of

Parameter corruption

Serial inclk

ElectrodeSwitchMatrixC

o ntr

o lle r

DAC#0

DAC#1

DAC#2

DAC#3

Analogout

LS

R2R

LSLS

Band-GapBias

REF

n Compatible with electrodematrix dimensionsw 4 x 4 electrodes (prototype)

M. Sawan24

Power supplyn External Coil, Rectifiers

Communicationn Downlink

w ASK: keep OOK for its simplicity(Tx/Rx), but need higher data rate

w «All digital» Demodulatorn Uplink

w Manchester LSK

System Controln Power managementn Communication Error

detection/correctionn Stimulator Instructions Dispatching

Monitoringn Actual Stimulating Currentn Voltage at every stimulation site

err flgcvtFSM

in / out reg

in/outFSM

errordetec. err

ctrl / status reg

clkLskclk

clkdata

LSKout

adc

o ut adc c lk a dc c vt

data(1)

data(N)

clk(1)

clk(N)

(to in/out FSM andctrl / status reg)Manch.

decod.envIn Envelop

detect.

PWRdata(1), clk(1)Vmon(1)err(1)

ASK in

LSK out

Monit.

Controller

...

PWRdata(N), clk(N)Vmon(N)err(N)

REF MREF

LSKmod

Env.det.

ClkDataext.

Regul

Clk

Data

Implant architectureInterface Module

M. Sawan

25

Coil Voltage

ASK in

VDDDownlink Tradeoffn High Q > Best efficiency, but

w Envelop transition speedlimited by Q-factor of receiver whenrectifier diodes voltage is reversed.

n … Limited Data Ratew Manchester encoding does not allow more

than ~500 kbps @ 13.56 MHz.

w Enhance Data Rate & Power Transfer

by reducing carrier OFF period.

n Digital decoderw Based on carrier signal attenuation

w Toggling of attenuated carrier detected byextra toggling of direct carrier.

Downlink CommunicationPower Efficiency vs Data Rate

Direct Coil

V -SS V j

V +DD Vj

Att. Coil VthL

VthH

Dig. I nDir

Dig. Att In

M. Sawan26

Monitoring & Data uplink

V / I monitoringV / I monitoringn Sampling at any programmable time

n Voltage measurementw Any stimulation site during normal stimulation

n Current monitoringw Actual stimulation current measurement.

Normal Stim.

Hi-Z

ADC

+-

enable

adc out

adc clkadc cvt

FromCtrler

ToCtrler

REF

I monit

Voltage Monit.

Co n

t ro l l

e rC

on t

ro l l

er

REF

REF

VD

DV

SS

DACDAC

DACDAC

R2R

R2R

Anal

og

Mon

it. B

us

REF

Hi-Z

Hi-Z

Current Monit.

M. Sawan27

Cortical stimulationStimulus mode/type

Anodic/CathodicStimulation

Anodic/CathodicStimulation

BipolarBipolarMonopolarMonopolar

Distant

reference

Distant

referenceDistributed

reference

Distributed

reference

Current

sink

Current

sinkCurrent

sink/source

Current

sink/source

Current

source

Current

source

– Flexibility• Modular: Variable number of stimulation sites

• Stimulation modes: Mono / Bipolar, Single / distributed return terminal

– Safety• Electrical Isolation: Allows stimulation current / voltage monitoring.

M. Sawan

4

28

Implementation Results

Stimulation Module (Stimulation Module (4 x 4 Matrix)• CMOS 0.18 µm, ~60 000 Gates

DownlinkDownlinknn > 1 Mbps @ 13.56 MHz, Duty Cycle = 67%> 1 Mbps @ 13.56 MHz, Duty Cycle = 67%nn 500 kbps @ 13.56 MHz, Duty Cycle > 85%500 kbps @ 13.56 MHz, Duty Cycle > 85%

Uplink : 200 kb/sUplink : 200 kb/s

Power: <1mW/SM @ 1MHzPower: <1mW/SM @ 1MHznn > 100 mW load capability; P (err) < 10> 100 mW load capability; P (err) < 10-6-6

Sufficient for 1000 stimulation sitesSufficient for 1000 stimulation sitesnn 256 stimulation patterns @ 50 Hz.256 stimulation patterns @ 50 Hz.

DownlinkImplantM. Sawan

CTRL

TEST structures

MONITORINGR2R AMP

ELECTRODESCONN / CTRL

DACs

BIAS

400 um400 um

29

Cortical stimulationImplant assembly

Modularn One controller &

several stimulationUnits

Flexible substrate

Power/data linkn Wireless

Data rate >1 Mb/sn Max. Stim. Freq. :> 1 kHz @ 500 Elec.

I / V Monitoring

Assemblyn Stim. Units.

Modularn One controller &

several stimulationUnits

Flexible substrate

Power/data linkn Wireless

Data rate >1 Mb/sn Max. Stim. Freq. :> 1 kHz @ 500 Elec.

I / V Monitoring

Assemblyn Stim. Units.

Physical Dimensionsn SM’s area : 3.5x3.5 mm2

n Substrate : 50 µm

n Stim Assembly: 1 mm thick

n Electrode Length: 2 mm

n IM’s Area : ~ 2 x 2 cm2

n Ctrler thickness : ~ 3 mm.

Physical Dimensionsn SM’s area : 3.5x3.5 mm2

n Substrate : 50 µm

n Stim Assembly: 1 mm thick

n Electrode Length: 2 mm

n IM’s Area : ~ 2 x 2 cm2

n Ctrler thickness : ~ 3 mm.

IM

SMs

M. Sawan 30

In vivo testingCollaboration with the INM

Subject

(Monkey)

Subject

(Monkey)

TimingTiming

PWR Data

Isolation

PWR Data

Isolation

StimulatorStimulator

OscilloscopeOscilloscopeComputer

Developmentboard

ParametersAmplitude, Pulse width,Inter-phase delay,Direction of first pulse,Pulse frequency, Pulsetrain duration, and Inter-train delay.

Only electrodes insubject; Configurablehardware; “Large” outputvoltage swing.

ParametersAmplitude, Pulse width,Inter-phase delay,Direction of first pulse,Pulse frequency, Pulsetrain duration, and Inter-train delay.

Only electrodes insubject; Configurablehardware; “Large” outputvoltage swing.

M. Sawan

Electrode matrix400 µm spacin;Platinum tips; Parylene-C coating, LowImpedance.

31

Acquired ImageAcquired Image

Image acquisition/processing

Challenges and Need for Optimizationn Stimulator

w Physical Properties Placement flexibility, wiringw Safety Chronic usagew Data rate High number of stimulation sitesw Stimulation Flexibility Prototypew Power efficiency High parallelism

n External Controlw Relevance of Data Low resolution

Actual Perceived ImageActual Perceived Image(Direct Fit)(Direct Fit)

Controlon phospheneparameters

Irregular- Density- Size- Sensitivity- …

M. Sawan32

Dedicated image processorn Allow to transform images over visuotopic maps;n Follow a bottom-up approach until percepts are understandable;n Must deal with every image encountered in real world;n Be safe for patient.

Required processingn Present image as the patient would see it;n Hardware implementation of processing algorithms;n Low power consumption topologies.

Minimize power consumption with other techniques than circuitoptimization approaches.

Image acquisition/processingMain specifications

M. Sawan33

Image acquisition/processingElectrically invoked phosphenes

w Visuotopic Mapsn Experimental determinationn Creation of random maps :

w Investigation of representation needed for a VCS.Original

Zones

EdgesM. Sawan

Constructed from experiments or fromVisual Data Base Generatorn Realistic Visuotopic Map

w Set of Phosphene Visual Field Coordinatesw Generated from probabilistic parameters

weighted with Schwartz retinotopic relationshipand estimated electrode placement.

n Probabilistic weighted PVFCvariations and phospheneparameters:w Sensitivity, Size, Brightness, etc.

34

Image acquisition/processingVisual Data Base (cont(cont’’d)d)

Enhancement/mappingn Image has to be enhanced for

clarity, then mapped to theavailable PVFCs.

n The Visual Field Emulator allowsperception prediction / evaluationfrom the research team.

Fit

Fitting example

M. Sawan

FoveaHorizontal Angle

Ver

tical

Ang

le

Sample Visuotopic Map

35

Image acquisition/processing

Power Optimization

average

average

With large electrode arrays, stimulation current represents a major partof power consumptionn Predetermined image scanning results in large current variations

n Adaptive scanning method regulates current demandw Equalize the required power and reduce the peak demands

Required stimulation current for 1 sample frame

Total stimulation current

Min. Current

Max. Current

AdaptiveScan

FixedScan

t

M. Sawan36

SummaryTypical topology of inductively coupled implantable electronic devicesDual-stimulator to recuperate the bladder functionsMain issues of a visual cortical stimulatorn Stimulation in the visual cortex may soon permit true artificial visionn Other undertaken research topics:

w Design of dedicated image sensorw High voltage stimulation techniquesw Modeling, simulation and in vivo validationw Measurement techniques of neuronal activitiesw Assembly several arrays

n 128, 500, and 1000+ microelectrodes.

Nanoelectronics will allow the design of additional devices.Microelectrodes & power sources are among the major works.Such medical devices include validation and involve physicians,engineers, health care professionals, families, etc.

M. Sawan