Robust Population Single Neuronal Calcium Signal Extraction Using SCOUT Allows for Longitudinal Analysis of Behavior-associated Neural Ensemble Dynamics

Abstract

In vivo calcium imaging enables simultaneous recording of large neuronal ensembles while engaged in operations such as learning and memory. However, such in vivo optical recordings are typically subject to motion artifact and background contamination from neurons and blood vessels. Further, population cell tracking across multiple recordings is complicated by non-rigid transformation induced by cell movements and field shifts. We introduce the novel method SCOUT for Single-Cell SpatiOtemporal LongitUdinal Tracking, consisting of two crucial parts: (1) imposition of spatial constraints on neuronal footprints extracted from individual optical recordings to improve ROI selection and eliminate false discoveries, and (2) application of a predictor-corrector, using spatiotemporal correlation of extracted neurons across sessions, for population cell tracking across multiple sessions. SCOUT empirically outperforms current methods for cell extraction and tracking in long-term multi-session imaging experiments across multiple brain regions. Application of this method allows for robust longitudinal analysis of contextual discrimination associated neural ensemble dynamics in the hippocampus up to 60 days.