neuropixels-analysis

You can run an end-to-end spike sorting pipeline by loading raw Neuropixels data to preprocess, apply drift correction, execute sorting, compute quality metrics, and export curated units. The workflow uses SpikeInterface and Python libraries like numpy and pandas.

Drift correction in neural data analysis is a preprocessing step that mitigates probe movement artifacts over time. It stabilizes spike locations before sorting, improving the accuracy of subsequent unit separation and quality metric calculations.

Yes, the workflow supports spike sorting execution using Kilosort4 and alternative sorters. It integrates with SpikeInterface to run modular sorting stages, allowing you to compare outputs and select the best sorter for your dataset.

AI-assisted curation integrates AI-driven visual analysis to classify units and inform merge decisions. It assists researchers in filtering noise and refining sorted units, maintaining human oversight throughout the quality metrics evaluation process.

Yes, this pipeline is designed for neuroscience researchers working with Neuropixels 1.0 and 2.0 data. It leverages SpikeInterface alongside pandas and matplotlib to load, preprocess, and analyze raw recordings within a Python environment.

AI-assisted curation requires human oversight to validate merge decisions and unit classifications. It serves as a visual analysis aid rather than a fully automated replacement, meaning researchers must manually review quality metrics to ensure accurate curation.