1

2

SOME RECENT DEVELOPMENTS IN THEDESIGN OF BIOPOTENTIAL AMPLIFIERS FOR

ENG RECORDING SYSTEMS

John Taylor, Delia Masanotti, Vipin Seetohul and Shiying Hao

Department of Electronic and Electrical EngineeringUniversity of Bath, UK

3

ACKNOWLEDGEMENTS

Professor N Donaldson, Department of Medical Physics & BioengineeringUniversity College, London, UK

Dr P Langlois, Department of Electronic and Electrical Engineering,University College, London, UK

Dr J Robbins, Department of Pharmacology, King’College,London, UK

Dr A Sapelkin, Department of Physics, Queen Mary University,London, UK

Dr R Rieger, Department of Electrical Engineering,National Sun Yat-sen University, Taiwan

Dr D Pal, Department of Electronics &Communication Engineering, B.I.T. Mesra, India

Dr M Schuettler, Department of Microsystems Engineering,University of Freiburg, Germany

Dr N Rijkoff Centre for Sensory Motor Interaction (SMI), University ofAalborg, Denmark

Engineering and Physical Sciences Research Council (UK) (EPSRC)

4

SUMMARY

The importance of real-time recording of electroneurogram (ENG)signals

The difficulties of achieving a stable interface between tissue andelectronic devices

Illustrations of current problems being studied at Bath University1 Velocity selective recording (VSR) of ENG signals2 In vitro recording of ENG from cloned neurons

Some possible future directions

5

THE IMPORTANCE OF ENG RECORDING

Although this area has been researched for some time, there isstill much demand for improved systems for real-time ENGrecording. Interest comes from eg:

•Neuroscientists requiring experimental data in fields such asneurophysiology and neuropharmacology

•Engineers requiring inputs for systems to control FunctionalElectrical Stimulation (FES) systems for a variety ofrehabilitation applications such as neurogenic urinaryincontinence by stimulation of the sacral roots

•There is a demand for recording methods with improvedfunctionality, eg. Velocity/diameter selective recording (VSR)

6

NERVE CUFFS FOR ENG RECORDING

electrodes

nerve

cuff

Avery and Wespic(1973)

Avery(1973)

Naples et al Kallesoe (1996)(1988)

Nerve cuff with tripolarelectrode assembly Various types of cuff design

7

MULTIELECTRODE CUFF (MEC)

ceramicadapter

cable

cuff

M Schuettler, 2006

•Polyimide thin-film technology

•Sputtered platinum electrodes

•Etched using oxygen plasma.

•The final MEC was 1.5 mm indiameter, 40 mm long andcarried eleven 0.5 mm wide,ring-shaped platinumelectrodes

8

RECORDING ENG USING NERVE CUFFS

Nerve

Cuff

Electrode

Amplifier

From tissue

To the central nervoussystem

+

-

This type of cuff/amplifier connectionis called a Quasi Tripole (QT). Itprovides good suppression of EMGand other artifacts. It only requires oneamplifier and is relatively simple toimplement. It has been much used inpractical ENG recording systems.

Output

9

10 CHANNEL ENG RECORDING SYSTEM

(N-2)

2nd-rankamplifiers

1st-rankamplifiers

nerve

electrode(rings)

insulatingcuff

adder

bandpassfilter

timedelays

output forone matchedvelocity

etc

(N-1)

(0)

(N-3)

AC couplingstage

subtractors

etc

cuff

Signal processing unit(SPU-digital)

digitisation

10

PREAMPLIFIER SCHEMATIC

VSS

VDD

Vref

Vout

Vin+ Vin-

34k

60k

50pFIbias

Q1 Q2

M1 M2

M4M5 M6

Q3

M7

M8M9

M10 M11

M12

M3

11

3mm

ACCoupling

Stage

ChannelSelection MUX

2nd rankAmps

Low-noise Pre-amps

10 CHANNEL ENG RECORDING SYSTEM

Die mounted in PGA packageASIC layout

12

EXPERIMENTAL SETUP FOR MEC RECORDINGS

13

XENOPUS FROG PREPARED FORREMOVAL OF SCIATIC NERVE

14

NERVE FITTED WITH MULTIELECTRODE CUFF INVITRO

15

MEASURED RESULTS IN FROG

Stimulation intensity 0.13 µC Stimulation intensity 1.01 µC

16

MEASURED VSR DATA IN FROG

Delay profiles corresponding totwo different stimulation intensities:

grey: 1.01 µC, white: 0.13 µC.

The bars have a width of 25 µs(reciprocal value of thesampling frequency)

17

BI-DIRECTIONAL INTERFACING OFELECTRONICS AND CULTURED NEURONS

•This is a collaborative programme involving E&EE at Bath, KCL/Guy’sHospital (London), Dept of EE at University College (London) and theDepartment of Physics at Queen Mary University (London)

•The overall aim is to enhance our understanding of how mammaliannerve cells can be connected optimally to integrated electronic circuitryfor neurobiological research and medical applications

•We intend to find an alternative method to traditional patch clampingwhich is non-intrusive, less labour intensive in use and is based on asimple, cheap, reproducible CMOS integrated circuit without any needfor elaborate post-processing

18

nucleus

axon

dendrite

cell body

terminal

0.2-20µm5-400µm

+

-

+

+

+

+

+

-

--

- +

-

-

+-

-

+-

++

-

+-

+

+-

+

-

+

+

+

-

-

+

+

culture medium

+ +

+

-

-

-

- cell membrane

sealing gap

cleft

sealing gap

+

NEURON-ELECTRODE INTERFACE: EQUIVALENT CIRCUIT

Si substrate

metallic pad insulating layer

CMOS circuitryextracellular signal

Vout

Cel

Rspr

Rel

Rseal

Csh

Rm

CRL

ENaEk EL

RkRNa

Vout

VM

gK gLgNa

Cel

19

PSPICE REALISATION OF THE HODGKIN-HUXLEY MODELOF A NEURON (1952)

VV

0

C71u

+-

G7

G

GAIN = 1

0.07

0

Vmv

-0.010613

VL

1000

POTASSIUM CHANNEL

0

R81e12

.0003

SODIUM CHANNEL

C81000u

R9

1e12

3PWR

25 0.1

EXP

1.000

-1

ILgL

-1PWRS

0.0555

V

EXP

4

Beta(m)

Alpha(m)

+-

G8

G

GAIN = 1

R10

1e12

V3

TD = 20m

TF = 1pPW = 20mPER = 30m

V1 = 0

TR = 1p

V2 = -100

1

V

0

0.036

C91000u

R11

1e12

0

30 0.1

LEAKAGE CHANNEL

EXP

0.05

1.000

gNa

-1PWRSEXP

Alpha(h)

Beta(h)

+-

G9

G

GAIN = 1

gK

V

.001

Vmv

ITotal

Vmv

n

m

Voltage Clamp

Cm

h

IK

.012

Vmv

-0.115

INa

VNa

0.12

C101000u

R12

1e12

4PWR

10 0.01

EXP

1.000

0.1 -1

-1PWRS

0.0125

EXP

0.125

Alpha(n)

Beta(n)

+-

G10

G

GAIN = 1

V

VKV

20

PAD-ONLY CHIP LAYOUT

Chip layout Packaged Die

21

GROWING NEURONAL CELLS ON CMOS

NG108-15 (neuroblastoma xglioma hybrid) cells, stainedwith methyl blue

Grown on CMOS microchipspre-coated with a cationicpolymer, (PE1), for 5 days

100μm

22

NEURONAL GROWTH ON POROUS SILICONEtched from polysilicon on a CMOS chip

23

ELECTRICAL IMPEDANCE SPECTROSCOPY (EIS) MEASUREMENTS ONPACKAGED CMOS CHIPS

10-1 100 101 102 103 104 105 106103

104

105

106

107

108

109

Frequency (Hz)

|Z|

pin1_chip1_31may06.zpin2_chip1_31may06.zpin3_chip1_31may06.zpin4_chip1_31may06.zpin5_chip1_31may06.z

10-1 100 101 102 103 104 105 106

-100

-75

-50

-25

0

25

Frequency (Hz)

thet

a

Packaged die with insulated bond wiresAveraged EISmeasurements

24

FUTURE WORK

•Neural Prostheses: (i) complete the demonstration of velocityselectivity (start January 2007) (ii) consider new methods of neuralrecording, e.g. using optoelectronics

•Electro-neural Interfacing: (i) Demonstrate capture of ENG fromsingle excited neurons; design on-chip signal processing foroptimum SNR. (ii) Complete modelling work

•SMART Orthopaedic Sensors: (i) Develop an optimal implantedsensor for hip micromotion detection and verify with in-vitroexperiments. (ii) Develop other possible applications for the use ofSMART sensors and actuators in modern orthopaedic applications

25

SYSTEM SPECIFICATION

15 nA, 20 nA, +,- inputs< 100 nAResidual DC input current

291 nV< 300 nVInput-referred rms voltage noise1 Hz -5 kHz

17 nV/√Hz, 1.5 nV/√Hz20 pA/√Hz, 2 pA/√HzInput-referred current noise density1 Hz, 1 kHz

11.5 nV/√Hz, 3.8 nV/√Hz20 nV/√Hz, 4 nV/√HzInput-referred voltage noise density1 Hz, 1 kHz

82 dB100 dBCMRR @ 1kHz

310 Hz, 3.3 kHz300 Hz, 3.5 kHzBandwidth

10,10010,000Midband gain

12 mm2Circuit area

24 mW< 50 mWPower consumption

±2.5 V±2.5 VPower supply

10Number of channels

MeasuredSpecificationParameter