Blind source separation based spike sorting for ephysiological recordings using large, densely packed multi-electrode arrays. Traces containing multiple units are first "unmixed" using a blind source separation method, Independent Component Analysis (ICA). Then the resulting components are evaluated on how successful the unmixing, or separation, has been. "Well-separated" components have their spikes extracted, to be used in later analyses as the set of spike times from a single neuron.
Code and documention in process of being made more user-friendly for publication and use. Some extraneous scripts remain. And some code for a proof-of-concept algorithm add on to handle burst spikes. A port of the code to Python is planned.
Paper will be posted on bioRxiv soon, a previous version is available in my dissertation: https://hdl.handle.net/2144/19751
Spike sorting code.
Primary function is RobustSpikeSort.m
Code for input options. Functionality has been replaced by Matlab's input parser, but code has not all been updated.