motion correction and noise threshold calculations

Open
#4,367 2 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Assessment

Difficulty
5/5
Estimated time
Over a week
Newbie friendliness
25/100
Issue type
Bug
Clarity
Needs clarification
Activity status
Stale
Tech stack
python
Domain
data

Research direction

Start by reviewing the linked Discord discussion and winter_2026_short.pdf for the reported motion-correction and channel-detection results. Reproduce the comparisons between 3-minute and 10-minute inputs, including motion estimates, noise vectors, peak counts, and dead-channel labels. Done means the discrepancies have documented causes and expected behavior.

Written by the indexing model from the issue text.

Description

motion correction preprocessing

I've shared a long thread of my motion correction results on the NPX discord

https://discord.com/channels/1446542289114763284/1468690639867740161

I've uploaded my powerpoint as a pdf here

In short I have three issues:

  1. I'm getting consistently poor motion correction results on what I believe to be a not-very-challenging input file
  2. For a 3 minute segment of data, the motion estimate changes, sometimes drastically, depending if the input is a 3 minute file or a 10 minute file. Small changes would make sense with how the motion vector is made-median, or potentially some causality effects, but some of these feels far outside what is reasonable.
  3. For the different data segments, the number of peaks used in motion correction appears very different. This is very likely due to different calculations of the MAD in the noise levels, which sets the peak thresholds for various methods. In my results the noise levels vector varies, but in the longer file values are sometimes as much as 20% higher than the shorter file. Still, this does not feel like the whole story.
  4. This feels separate, but is possibly related. My detection of bad channels seems inaccurate. In both probes numerous seemingly-good channels are called dead, and in one probe the 191 reference channel is not labeled dead.

winter_2026_short.pdf

Dominant language
Python
Stars
847
Forks
280
Avg merge
3d 9h
Merged PRs (30d)
29

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from SpikeInterface/spikeinterface

All issues in SpikeInterface/spikeinterface

Similar issues

More Python issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.