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

Area of Science:

• Systems Neuroscience
• Computational Neuroscience
• Neurotechnology

Background:

• High-density microelectrode arrays (MEAs) are crucial for systems neuroscience but face challenges from brain tissue motion relative to the array.
• Accurate registration of electrophysiology data is essential for reliable downstream analysis, especially in human recordings.

Approach:

• Introduced DREDge (Decentralized Registration of Electrophysiology Data), a robust algorithm for registering noisy, nonstationary extracellular electrophysiology recordings.
• DREDge estimates motion from action potential (AP) frequency band spikes and enables high-temporal-resolution motion tracking in the local field potential (LFP) band.
• Implemented DREDge for human intraoperative recordings, deep probe insertions in nonhuman primates, acute mouse recordings, and chronic mouse implantations.

Key Points:

• DREDge correction in the LFP band recovered evoked potentials and reduced spike sorting errors in human recordings with fast motion.
• Enabled tracking of centimeter-scale probe motion across brain regions in nonhuman primates.
• Demonstrated improved motion correction in acute and chronic mouse recordings, including those with ultra-high density probes and long-term drift.

Conclusions:

• DREDge provides automated, scalable registration of electrophysiological data across species, probe types, and drift scenarios.
• Enables stable, high-fidelity neural data for advanced scientific analyses.
• Facilitates more reliable interpretation of neural activity in diverse experimental settings.