LightSuite: scalable atlas registration and cell detection across the central nervous system

Abstract

Deciphering brain-wide circuits requires large animal cohorts, yet computational bottlenecks in atlas registration and whole-volume cell detection reduce precision, reproducibility, and sample yield. We present LightSuite, an open-source framework enabling scalable, multimodal registration and cellular mapping across the central nervous system (CNS). LightSuite combines damage-resilient 3D point-cloud initialization with user-steerable non-rigid registration, and GPU-accelerated local background normalization for rapid, illumination-invariant cell detection. Applying LightSuite to Fos-dependent activity labeling in mice, we extracted reproducible, layer-resolved profiles of active cell density, soma-size distributions and per-animal labeling-efficiency factors. LightSuite computationally untwists and registers cleared spinal cords to map myeloarchitecture, reconstructs serial wide-field histology to map surface GluA1 distributions, resolves optic fiber tracks, and aligns in vivo functional ultrasound hemodynamics into common atlas coordinates. LightSuite establishes an integrated, accessible platform for cohort-level structural, molecular, and functional CNS mapping.