In Kürze

3. Eingebettete und interaktive Systeme

Fonds HES-SO

Roland Scherwey
Kompetenzinventar

April 2017 - September 2018

The project is developing a platform for efficient processing and storage of neuronal spike activities.

The last decade has witnessed a renewed interest for in vitro approaches in the fields of drug discovery and toxicity testing. One very promising approach is the use of engineered human neural tissue fabricated from iPS cells. To ensure the rapidity of the tests in neural tissues the traditional chemical, cytosolic and histologic read-outs have been replaced by an electrophysiological read-out, i.e. the recording of the electrical activity of neurons using smart Petri dishes that incorporate electrode arrays. A big challenge for recording such activity is the very large amount of data generated by the electrode arrays, resulting in cumbersome and long data analysis to be performed in order to get the final experimental results, and the reliability of the systems. Within this context, the aim of the SpikeOnChip project is to develop a platform for efficient processing and storage of neuronal spike activities. It will not only offer recording of observed electrodes (up to 64), but also an on-chip analysis that will allow to reduce the quantity of data by a factor of 20x, by selecting only the interesting portions.

The platform will be based on a ZedBoard embedding a Zynq (FPGA + ARM processor in one die). This data reduction will allow to save memory, and as a result allow the system to run autonomously for a longer period of time. Connection will be supplied by a Wifi link in order to transfer data to a PC, however gaining autonomy will imply a more reliable system in the sense that if the connection is lost at some point, data can be stored for up to 3 hours without losing any important information. Data acquisition will be done by an Intan chip that translates raw voltages from the electrode array onto serial digital data that is sent to an FPGA. The processing performed on-chip will consist in signal filtering of the raw data, spike detection, noise rejection, and field potential frequency analysis.
Finally a software will also be developed so as to let a user control the embedded platform and to visualize the analysis results. This project will imply a validation stage, performed by biologists, to ensure the system not only works with respect to the specifications, but is also usable by end users.