Design And Implementation Of 16-channel Real-time And Unsupervised Neuronal Spike Sorter Chip

Damage to the hippocampus will result in the loss of ability to form new long-term memories and cognitive disorders.Hippocampus is also one of the first regions to suffer damage.At present,there is no effective medical treatment for this issue.Hippocampal cognitive prosthesis is proposed to replace damaged regions of the hippocampus to mimic the function of original biological tissue.This prosthesis requires a spike sorter to detect and classify spikes in the recorded neural signal.A 16-channel real-time and unsupervised spike sorting processor implemented on 40 nm CMOS process is presented in this paper.During the design,the accuracy and computational complexity of three spike detection algorithms are compared comprehensively.An automatic threshold estimation method suitable for hardware implementation is proposed for the Osort clustering algorithm,which is demonstrated to be more robust and converge faster.Besides,it consumes less hardware resources.Bayes optimal template matching(BOTM)classification algorithm is optimized into PBOTM(Preselection BOTM),which can detect and classify partially overlapping spikes.Compared with BOTM,the computational complexity of PBOTM reduces by 99.15%in average,but the accuracy only reduces by 0.3%.When implemented on hardware,power consumption of this module reduces by 84.4%and area reduces by 13.75%(the results have been published in a SCI journal).The test result shows that power consumption of this chip is 304.7 ?W(19.0 ?W/ch)and core area is 0.281 mm2(0.0175 mm2/ch).The detection accuracy is 98.3%and classification accuracy is 93.4%.Compared with the other four spike sorting processors,the proposed chip achieves a relatively high detection and classification accuracy with a small area.It also has the ability to deal with partially overlapping spikes,which is not reported in the other work.The realization of this chip can not only promote the development of hippocampal prosthesis,but also can be applied to various neuroscience experiments for online spike sorting.