neuropixels-analysis

Run spike sorting and drift correction on Neuropixels data by loading raw electrophysiology recordings to apply preprocessing, motion correction, and automated sorting, producing curated neural units for downstream analysis.

AI-assisted curation for spike sorting applies automated quality metrics and visual analysis prompts to refine sorted units, using Allen, IBL, or Strict criteria to filter noise and produce publication-ready neural outputs.

SpikeInterface supports GPU-accelerated sorters like kilosort4 as well as CPU fallbacks including spykingcircus and mountainsort5, enabling flexible spike sorting across different hardware environments for Neuropixels analysis.

Export curated spike sorting results to Phy or NWB format by running the end-to-end Neuropixels pipeline, which generates ready-to-share visualization files and standardized neural data outputs for publication.

Yes, you can perform bad-channel removal and common median reference on extracellular recordings, as the pipeline applies standardized preprocessing steps including phase correction and motion correction to prepare Neuropixels data.

Automated curation limitations stem from the strictness of Allen, IBL, or Strict criteria applications, which may require manual visual verification in Phy to ensure no true units are incorrectly discarded during the spike sorting process.