INTENSE: Detecting and disentangling neuronal selectivity in calcium imaging data

A research team led by Nikita Pospelov from Moscow State University has published INTENSE (INformation-Theoretic Evaluation of Neuronal SElectivity), a breakthrough computational framework for analyzing calcium imaging data from freely behaving animals. The tool addresses a critical bottleneck in neuroscience: while calcium imaging can record activity from thousands of neurons simultaneously, traditional methods struggle to accurately link this neural activity to specific behaviors due to temporal autocorrelations and calcium indicator kinetics. INTENSE solves this by applying information theory—specifically mutual information calculations—to detect statistically robust associations between raw fluorescence signals and behavioral variables like place, head direction, and locomotion.

INTENSE introduces two key innovations: circular-shift permutation testing that preserves temporal structure while controlling false discoveries, and conditional mutual information-based disentanglement that separates genuine mixed selectivity from associations driven by behavioral covariance. Validated on synthetic datasets, the framework demonstrated 90% detection accuracy across varying signal-to-noise ratios, significantly outperforming methods lacking temporal controls. When applied to real CA1 hippocampal recordings from mice exploring open fields, INTENSE successfully mapped neurons with selectivity to multiple behavioral variables while distinguishing redundant from genuinely multivariable encoding. The open-source tool now enables neuroscientists to perform high-throughput, information-theoretic mapping of brain-behavior relationships, accelerating the translation of large-scale recordings into testable circuit-level hypotheses about cognition and memory formation.