B-Spline vs. Lifting Architecture for Neuroprosthetic Devices

Advances in high-density microelectrode arrays has created a need for high level signal processing to cope with the immense data throughput. The discrete wavelet transform (DWT) has been shown to optimally reduce the amount of data throughput, while concisely preserving the information in the data. Two factorizations, lifting and B-spline, have been proposed for implementing the DWT in hardware. A comparison based on critical path, memory requirements, and computational hardware is drawn. The context of this comparison focuses on multichannel and multiresolution neuroprosthetic devices