Automated Virtual Pathology Panels for ML-Assisted Interpretation of Mass Spectrometry Imaging

Abstract

Mass spectrometry imaging (MSI) records rich molecular spectra at each pixel, but pathology-oriented interpretation benefits from complementary views of the same tissue, loosely analogous to the multiple stains used in histopathology. We present an expert-aligned framework for constructing multi-view MSI panels. Soft Landmark Contrast Edges (SoLaCE) extracts molecular boundaries directly from high-dimensional spectra. Because standard structure-preservation metrics aligned poorly with rankings by pathologists, we combine luminance contrast and chromatic diversity with SpecEdgeDice, a boundary-aware measure of agreement between visualization edges and SoLaCE boundaries. Parametric MiCS+LMC (pMiCS) uses a neural network trained on subsampled data to distill multiple MSI segmentations into a reusable spectral-to-RGB mapping, enabling rapid full-image inference, out-of-sample projection, and more consistent color semantics across aligned images. A concept-based interpretation procedure explains pMiCS outputs through sparse mixtures of spectral concepts, scoped as a model output-equivalent summary rather than a molecular identification. We integrate these components into Virtual Pathology Panels, which use hyperparameter optimization to select fast, high-performing or spatially complementary views. This framework supports future workflows that combine automated, AI/ML-supported, morphology-oriented tissue assessment and molecular analysis within a single MSI acquisition.