ROBUST ONLINE MULTIBAND DRIFT ESTIMATION IN ELECTROPHYSIOLOGY DATA.

Area of Science:

• Systems Neuroscience
• Neurotechnology
• Computational Neuroscience

Background:

• High-density electrophysiology probes are crucial for systems neuroscience research.
• Probe motion artifacts complicate data analysis, especially in human recordings.
• Existing motion tracking methods have limitations in temporal resolution and scalability.

Purpose of the Study:

• To improve the state-of-the-art in tracking electrophysiology probe motion.
• To enable scalable and automated registration of challenging electrophysiology datasets.
• To enhance the robustness and temporal resolution of motion tracking.

Main Methods:

• Extended decentralized methods using multiband local field potential (LFP) information alongside spikes.
• Developed an efficient online motion tracking algorithm for scalability.
• Introduced a structure-aware objective and adaptive parameter selection for improved robustness.

Main Results:

• Achieved sub-second temporal resolution for probe motion registration using LFP data.
• Demonstrated the scalability of the online algorithm for long and high-resolution recordings.
• Enhanced the robustness of motion tracking for challenging human and mouse datasets.

Conclusions:

• The novel approach enables fully automated and scalable registration of electrophysiology probe motion.
• This advancement facilitates more accurate downstream analyses in systems neuroscience.
• The method holds potential for real-time applications in neurotechnology.