Evaluating the influence of the bit error rate on the information of neural spike signals

Abstract

This paper analyzes the influence of the bit error rate (BER) on the information in neural spike signals. State-of-the-art implanted neural recorders use an array of more than 100 electrodes to acquire neural signals. For optimal data analysis at the receiver and maximal mobility of the person wearing the implant, the entire neural spike signals need to be sent wirelessly to the receiver. In general, wireless data transmission is subject to higher BER than wire-bound data transmission. For wireless data transmission the influence of the BER on the information loss in the data needs to be examined to optimize the transmitter in terms of power consumption. This paper uses the spike sorting program OSort for an evaluation to which extent the information in neural signals is preserved after the distortion by random bit errors. Consequently, not only the number of incorrectly detected spikes is investigated, but also the change in the waveforms of the spikes. Synthesized and recorded neural data sets are utilized for the BER evaluation. A BER lower than 2×10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-6</sup> is found to be sufficient to leave the spike information of the recorded data sets virtually unchanged.