Back to the projects list

Incorporate motion correction techniques into SpkeInterface

Key Investigators

• Alessio Buccino
• Samuel Garcia
• Ben Dichter

Project Description

Motion of the probe has been a major impediment to high quality spike sorting on high density probes such as the Neuropixel probe. Recent developments in preprocessing have displayed promising results for solving this issue, but these algorithms are varied, complex, and multi-step. Here, we are incorporating these motion correction methods into SpikeInterface so that they are easier for users to access. The motion correction process can be summarized in the following steps:

Using action potentials:

1. detect peak
2. localize peaks
   1. center of mass
   2. monopolar triangulation
3. motion estimation
   1. rigid or non-rigid
   2. decentralize
   3. motion cloud

Using LFP: …

Objectives

Incorporate the following methods into SpikeInterface

• detect peaks
• localize peaks/center of mass
• localize peaks/monopolar triangulation
• motion estimation/rigid decentralize
• motion estimation/non-rigid decentralize

Each of these methods requires implementation, testing, and documentation of usage.

Approach and Plan

1. Create/refine motion correction map here.

Progress and Next Steps

Motion correction map: https://docs.google.com/drawings/d/1igJuCqULG7lNq4TfOSNr351Fm03lbeddXLA9Ev77bTA/edit

Materials

• https://spikeinterface.github.io/blog/spikeinterface-motion-estimation/

Background and References

See initial progress here