DREDge: robust motion correction for high-density extracellular recordings across species

Area of Science:

• Systems Neuroscience
• Computational Neuroscience
• Neurotechnology

Background:

• High-density microelectrode arrays are crucial for systems neuroscience.
• Brain motion relative to these arrays complicates data analysis.
• Existing methods struggle with noisy and nonstationary electrophysiology data.

Purpose of the Study:

• To introduce DREDge (Decentralized Registration of Electrophysiology Data), a novel algorithm for robust motion registration.
• To enable automated, high-temporal-resolution motion tracking using local field potential data.
• To provide a foundation for scalable, automated registration of electrophysiological data.

Main Methods:

• Developed DREDge algorithm for decentralized registration of electrophysiology data.
• Utilized action potential and local field potential data for motion estimation.
• Validated DREDge on human intraoperative, nonhuman primate, and mouse (acute and chronic) recordings.

Main Results:

• DREDge reliably recovered evoked potentials and single-unit spike sorting in human recordings.
• Tracked motion across centimeters and multiple brain regions in nonhuman primates.
• Significantly improved motion correction in acute mouse recordings, particularly with ultrahigh-density probes.
• Achieved stable motion tracking in chronic mouse implants despite neural activity changes.

Conclusions:

• DREDge offers automated, scalable motion registration for electrophysiological data.
• The algorithm is effective across species, probe types, and drift conditions.
• Enables more robust downstream analyses of complex neural datasets.