WTEC Brain Computer Interface (BCI) Workshop: Sensor Technology

Greg A. GerhardtUniversity of Kentucky Health Sciences CenterDepartments of Anatomy and Neurobiology,

Neurology and Psychiatry

WTEC Workshop on Brain Computer Interface Research: 21 July 2006 Sponsors: NSF, TATRC, NIBIB, NINDS, DoED

Sensors in BCI – Study Highlights

• Science of BCI in North America and Europe

• The majority of BCI science in North America involves “invasive” technologies, i.e., multi-electrode recordings from arrays of electrodes implanted directly into brain.

• However, certain BCI sites in Europe are capable of providing technologies that could aid in the advancement of “invasive” sensor technologies. These sites could be an untapped resource!

• The majority of BCI science in Europe involves “non-invasive” technologies, i.e., multi-electrode recordings from arrays of electrodes mounted onto the surface of the skull.

Sensors in BCI – Definitions

• Invasive Technologies – wire arrays, Electrocorticographic (ECoG) strips, microfabricated electrode arrays (MEAs)

• Non-invasive Technologies – EEG, “head-ware devices”

• Enabling Technologies – In Vitro technologies such as MEAs

Initial Work with Electrodes (pre-1965)

• Hess (1932) - first to implant electrodes in diencephalon of cat

• Fischer (1957) - various metals/insulators used as single wire electrodes; 1-2 mm injury around tract

• Collias (1957) - Histopathological analysis; evolving response; astrocyte capsule formation by 1 mo.; FBR to electrode

• Delgado (1961) • Robinson and - Reinforced histological findings

Johnson (1961) **Courtesy of Patrick Tresco

Evolution of Electrode DesignsMICROWIRES

• Salcman and Bak (1973) - Record with parylene-coated microwires• Woodward

and Chapin (1980s) - Developed multi-wire arrays-------------------------------------------------------------------------------SILICON MICROELECTRODE ARRAYS

• Wise and Angell (1970, 1975) - Use IC technology to develop

microelectrodes• BeMent (1986) - Developed first multi-site electrode

from Si (Michigan-style electrode)----------------------------------------------------------------------------------• Campbell (1991) - Developed first monolithic multi-shank

electrode from Si (Utah Electrode Array)

**Courtesy of Patrick Tresco

ELECTRODE ARRAY

19

CA1

CA3

DG

16 8

CA3 CA1

S

TLATERAL

MEDIAL

NOSEPOKE

R-SAMPLEL-NONMATCH

L-SAMPLER-NONMATCH

10

11

12

13

16

1

2

5

6

7

CA1

CA3

0 50 100 150 Time (sec)

5

0

Num

ber o

f spi

kes

5

0

Num

ber o

f spi

kes

Neural Activity = Vector in N-dimensional space

i,tX X = Firing Rate, i = Neuron, t = time

“Micro-wires” –the work horse sensors of many multi-single unit recording labs

Courtesy of Drs. Sam Deadwyler and Rob Hampson

Micro-wire Recordings of Single-Unit Activity

Courtesy of Scientific American and John Chapin

“Michigan” Probes

Wise, et al.(2004), Proceedings of the IEEE.Hetke and. Anderson (2002). Handbook of Neuroprosthetic Methods.

4/11/2003

Michigan Probes as a ‘Toolkit’Basic probe assembly for chronic studies in animals

Kipke et al. (2003). IEEE Trans Neural Systems and Rehab. Engin.Vetter, Kipke, et al. (2004) IEEE Trans Biomed Eng

Cortical controlCortical control~60 functional channels~90 high-quality spikes

Discriminated spike waveforms

Spike rasters for acquired robot control

Schwartz et al. (U Pitt.)Kipke et al., (Univ. of Michigan)

4/11/2003

Microfabricated Parylene Probes

Chronic unit recordings(FP5, day 7)

500 µV

-500 µV

0

Microscale drug-delivery

Courtesy of Daryl Kipke, Univ. of Michigan

Future Wireless Technologies (Kipke et al., Univ. of Michigan)

Direct communication with the CNS: The ‘Utah Electrode Array’.

• MEM’s built silicon microsystem.

• 100 electrodes.• Each electrode

communicates with 2-3 neurons.

Courtesy of John Donoghue and Cyberkinetics

CNS Interconnect Systems

Courtesy of John Donoghueand Cyberkinetics

Neuron-Silicon Communication:Conformal Multi-Site Recording Electrode Arrays

Trisynaptic conformal designaligned with rat acute slices

DG

CA3

CA1

100 µm

Designed to allow recording from DG, CA1, and CA3 simultaneously

Designed for external single site stimulation

Capable of multi-site internal stimulation

Precisely aligned with averaged hippocampal slice cytoarchitecturalcoordinates

Courtesy of Ted Berger, USC

“Ceramic-based Microarrays”

Side-by-Side

Serial

Ceramic-Based Conformal Microelectrodes

Unique Features of Ceramic-based “Conformal” Microelectrodes

1. Ceramic (Al2O3 ) substrates 37.5 to 125 µm

2. Long electrode configurations (1-20 cm)

3. “Multi-purpose” tip and shank designs

W3

15x333 μm

1350 μm

W2

20x150 μm

600 μm

USC, Wake Forest and Univ. of Kentucky

MEAS with Flexible Connectors Analogous to Subdural Designs

Spencer-Gerhardt Microarray (Chemistry and Physiology)

SG-1 prototype

Spencer-Gerhardt (SG-1)

microelectrodes

Electrode tip

Ceramic microelectrode

SG-1

Sub-dural strips

Univ. of Kentucky and Ad-Tech Medical Instruments

Major Areas of Research

• What factors improve longevity of the recordings?

• Failure analysis of components over 1-12 month periods.

• How long do current designs last?• How do we develop designs that last for

ca. 5-10 years?

Electrocorticographic (ECoG) Control of Brain Computer

Interfaces

Human ECoG Grids for Epilepsy

ECoG: Anatomical localization: Albert-LudwigsUniversity, Freiburg, Germany

L

H G F E D

C B A

12345678

FL

TL

Development of Epidural Micro ECoG Grids

Courtesy of Dan Moran, Washington Univ., St. Louis

In Vitro MEA’s

Multi Channel Systems In Vitro MEAS (Reutlingen, Germany)

60 channel arrays

Professor Peter FromherzMax Planck Institute for Biochemistry, Munich, Germany

Rat neuron on electrolyte-oxide-silicon (EOS) field effect transistor. a) Electron micrographs (colorized) of a hippocampal neuron on a silicon chip array; b) Schematic cross section of a neuron on a buried-channel field-effect transistor with blow-up (drawn to scale) of the contact area.

16,384 Element Silicon-Neuron Array Recordings

Cultured Hippocampal Slices

7.4 μ resolution – 2 KHz Measures

Max Planck Institute for Biochemistry, Munich, Germany

Non-Invasive EEG-Based BCI

Brain-Computer Interface (BCI)Brain-Computer Interface (BCI)

BCI

Applicationclosedloop

system

visual feedback

brain signal control signal

Mobile EEG system from g®.tec (Austria)

Sensors - Non-Invasive BCI

• Need for “dry electrode” systems

• More “on electrode” electronics for improved signal-to-noise

Journal of Neuroscience Methods Special Issue on BCI

• Editors – Ted Berger and Greg Gerhardt

• Manuscripts due by 12/1/06

• 1-2 volumes + overview - Currently 12 tentative manuscripts