MEDiCINe

MEDiCINe estimates motion in neural electrophysiology recordings to improve spike sorting accuracy by correcting for movement in laminar multielectrode arrays.

Key Features:

• Motion Estimation by Distributional Contrastive Inference: A novel motion estimation method that addresses the challenge of robust and accurate motion estimation due to the stochasticity of neural data.
• Performance on Simulated Recordings: Outperforms existing motion estimation methods on an extensive suite of simulated neurophysiology recordings.
• Validated on Primate and Rodent Recordings: Accurately estimates motion in primate and rodent electrophysiology recordings with a variety of motion and stability statistics.
• Spike Sorting Integration: Serves as a preprocessing step for standard spike-sorting methods and leads to more accurate spike sorting results.

Scientific Applications:

• Spike Sorting Improvement: Corrects for motion caused by pulsation, changes in intracranial pressure, and decompression of neural tissue after insertion to stabilize recordings and enable identification and tracking of single neurons across time.
• Cross-Species Electrophysiology Analysis: Applicable to primate and rodent electrophysiology datasets with diverse motion and stability characteristics.

Methodology:

MEDiCINe uses distributional contrastive inference for motion estimation in neural electrophysiology recordings.