Neural Spike Compression Based On Split Vector Quantization For Implantable Bmis

This paper reports a novel spike compression approach for implantable intra-cortical neural recording microsystems based on Split Vector Quantization (SVQ). The proposed method presents a spike compression ratio 14.8 at the cost of classification accuracy (CA). The average value of CA is 94% over a wide range (7 to 15) of signal to noise ratios (SNR) of the neural signal.