The brain’s lipid landscapes uncovered by mass spectrometry imaging and explainable machine learning

Recent computational advances in mass-spectrometry imaging (MSI) now enable unprecedented insight into organ-wide molecular composition and functional architecture. We present the first high-resolution, molecular-computational atlas of the brain lipidome, covering 123 anatomically defined regions with 191 polygonal annotations derived solely from MSI data—no auxiliary imaging required. To overcome annotation ambiguity and MSI complexity, we introduced the Computational Brain Lipid Atlas (CBLA), a graph-based visual-explainability framework that generates virtual landscape visualizations (VLV) of lipid distributions across brain substructures. CBLA integrates dimensionality reduction and ensembles of supervised models to (i) refine annotations, (ii) elucidate inter-regional relationships, (iii) interpret model behavior, and (iv) formulate biologically testable hypotheses. Applying CBLA revealed novel lipid-distribution patterns, functional integrations, anatomical connectivities, and region-specific disease signatures. A new algorithm decomposes annotated regions into precise m/z features and resolves full-precision m/z values from binned MSI data, producing a comprehensive, high-resolution brain lipidome map. This resource underpins downstream studies, exemplified here by characterizing the molecular makeup of Aβ plaques, their spatial arrangement, and their connections to surrounding tissue.