KiloSort
KiloSort copied to clipboard
discard artifacts
For example as brought up in #98. This could either be a pre-processing step on the raw data, or a step during template matching to ensure spikes are not matched if they have too much variance orthogonal to the template.
Any other ideas welcome.
What should our expectation be for any further development of kilosort?
It seems like many fundamental aspects/issues (such as this one) have been left totally unfinished. Not meaning to take a jab at the creators (...people get busy, projects/funding move on, etc), just trying to consider whether its wise for us to build internal infrastructure around it.
Artifact removal is one of the most modular parts of the pipeline, so I encourage anyone who wants to write that code and contribute it.
However, I encourage users to build their own artifact removal preprocessing, because the artifacts can be very different from lab to lab. There is no all-encompassing artifact removal step that will fix everyone's problems, but perhaps there could be multiple variants that users could choose from.