Design of basic block of neural signal detection chip

Abstract

Vagus nerve stimulation, an emerging technology for the treatment of many neurological diseases, is considered an alternative to medical treatment and surgical treatment. This paper introduces the circuit implementation of the SVM (Support Vector Machine) module in the neural signal monitoring chip classifier. The support vector machine module includes an exponential function operation unit and a scaling unit. The exponential function operation circuit is designed based on the coordinate rotation digital calculation method (CORDIC). In order to reduce the amount of computation, some of unimportant or redundant features are removed from the complete feature set by selecting the patient's EEG (electroencephalogram) signal path. In addition, the decision function is transformed to optimize the transmission path of the data. Finally, the indicators for evaluating this system are given by the simulation results.