Neural Data Compression with Wavelet Transform: A Vocabulary Based Approach

An algorithmic approach to develop the vocabulary of the nervous system and to use the vocabulary to communicate with the outside world is presented. The vocabulary is constructed using wavelet analysis of the recorded waveforms. Spikes of different frequency and amplitude from different channels are identified to construct unique signatures and relate them to physiological activities. A vocabulary-based communication of recorded action potentials renders two major advantages: a) it allows transmission of recorded data with large compression, thus, saving power and communication bandwidth of the integrated telemetry device; b) it helps easy mapping of alphabets in the vocabulary to muscular dynamics, which facilitates micro-stimulation based neural prostheses. In this work, we study the effectiveness of the proposed approach in neural data compression. Simulation results on pre-recorded data from the buccal nerves of a sea-slug shows that the proposed approach results in up to 80X compression