| |
|
| [ title ] |
 |
Spike Discrimination for a Hybrid Bio-CMOS Chip |
[ abstract ]
 |
|
Goal: The overall goal of INPRO: Information Processing for Natural Neural Networks is to establish a novel method of information processing by combining natural neurons with silicon technology. This will be pursued by (a) realizing a prototype of a bioelectronic circuit, and (b) establishing its information processing capabilities.
The hybrid bio-CMOS chip for simultaneous recording and stimulation has been developed at the Physical Electronics Laboratory. My job in this context was the evaluation of spike discrimination methods in the more general field of pattern recognition, and to carry out some preliminary tests with suitable methods.
Results: Based on the literature available at that time with respect to information processing in natural neural networks, the effectiveness and necessity of spike sorting had to be put into question. As a matter of fact, no algorithms to date can deal with the overlaps in multi-unit recordings - resulting in misclassifications and thus a distortion of stochastic properties of the neural activity. While relatively clear signals can be recorded in-vitro, where the environment can be observed and controlled to some degree, there is no such solution to date for the ever changing properties of cells and the interface itself in-vivo. |
| [ location ] |
|
Swiss Federal Institute of Technology, Institute of Quantum Electronics, Physical Electronics Laboratory, Zurich, Switzerland |
| [ framework ] |
|
Semester Thesis II |
| [ duration ] |
|
April - July 2003 |
|
 |
|
|
