ABSTRACT:

Recent advancements in neural recording techniques have enabled neuroscientists to monitor cell activity from a large population of neurons at a certain brain region of freely behaving mice or rats in vivo. However, real-time neural signal processing for closed-loop feedback applications, especially under strict hardware and energy consumption requirements, remains a challenge. In this talk, I will introduce our research work on real-time EEG and calcium image processing through customized algorithm and accelerator designs on the field-programmable gate array (FPGA). In the first part, I will introduce an LSTM-based method to approximate IIR/FIR filters and the corresponding accelerator and circuits design for real-time EEG signal processing. In the second part, I will describe customized algorithms and accelerators to enable real-time calcium image motion correction, enhancement and trace extraction with sub-ms latency. The proposed technique opens up possibilities for real-time neural signal processing and decoding for a range of closed-loop feedback applications.

BIO:

Zhe Chen conducts postdoctoral research under supervision of Prof. Jason Cong from Computer Science Department and Prof. H. Tad Blair from the Department of Psychology. His research interests are in closed-loop systems for neuroscience, reconfigurable and customized computing for domain specific applications. He was a recipient of the UCLA Chancellor’s Award for Postdoctoral Research in 2019, and a recipient of the UCLA Depression Grand Challenge Fellowship in 2016. His research works received the Best Paper Award in ISLPED’18 and the Excellent Student Paper Award in ASICON’13. His work also received Pearl Cohen 2nd Place Winner Poster Award on the UCLA Bioscience Innovation Day in 2019.

Hosted by Professor Jason Cong

Location: Via Zoom Webinar