About a fixed noise - Githubissues

csn-le / wave_clus

A fast and unsupervised algorithm for spike detection and sorting using wavelets and super-paramagnetic clustering

127 stars 66 forks source link

About a fixed noise #105

Closed koro30 closed 5 years ago

koro30 commented 6 years ago

Hi Fernando, I have some human intracraneal recordings (from midbrain) and there is a constant noise that appears every 20ms. Fortunately most of the time waveclus isolates it as a different cluster, and is ok because all the noise is isolated in one. But many times this 20ms contaminates the other cluster and I don't know how to get rid of it. It looks like a tonic firing but is too clean to be it. Do you know if there is a way to do it? Thanks!

ferchaure commented 6 years ago

Hi, I don't have a perfect solution. You could:

Manually select and clean the clusters using the gui.
using a smaller value for the max distance in the template matching step (if the unforced cluster doesn't mix noise and spike)
if the noise is more or less always the same , maybe you can code a comparator to remove all the spikes that have a small euclidean distance to the noise after the detection in the _spikes.mat file