Robust Online Multiband Drift Estimation in Electrophysiology Data

Abstract

ABSTRACTHigh-density electrophysiology probes have opened new possibilities for systems neuroscience in human and non-human animals, but probe motion (or drift) while recording poses a challenge for downstream analyses, particularly in human recordings. Here, we improve on the state of the art for tracking this drift with an algorithm termedDREDge(DecentralizedRegistration ofElectrophysiologyData) with four major contributions. First, we extend previous decentralized methods to exploitmultibandinformation, leveraging the local field potential (LFP), in addition to spikes detected from the action potentials (AP). Second, we show that the LFP-based approach enables registration atsub-secondtemporal resolution. Third, we introduce an efficientonlinemotion tracking algorithm, allowing the method to scale up to longer and higher spatial resolution recordings, which could facilitate real-time applications. Finally, we improve therobustnessof the approach by accounting for the nonstationarities that occur in real data and by automating parameter selection. Together, these advances enable fully automated scalable registration of challenging datasets from both humans and mice.