Enzlein, et. al., 2024, extracted plaque data for machine learning

Utilizing multimodal mass spectrometry imaging (MSI) combined with machine learning techniques, this study investigates the molecular heterogeneity of amyloid-β (Aβ) plaques and associated lipid profiles in post-mortem brain samples from Alzheimer’s disease (AD) and amyloid-positive cognitively unaffected (AP-CU) individuals. Our analytical approach permitted single-plaque level investigation, studying many individual plaques, instead of focusing on a population mean, which could hide potential sub-populations, revealing distinct populations of amyloid plaques characterized by differential Aβ and lipid compositions. Notably, the integration of MSI data with machine learning based feature extraction enabled the identification of Aβ38 and ganglioside GM1 as significant molecular markers differentiating AD from AP-CU pathology. These findings suggest that the heterogeneity in Aβ metabolism and lipid homeostasis, as revealed through precise analysis, is a key factor in the pathogenesis of AD and implies that total amyloid burden alone is an insufficient marker for the disease. The application of MSI and machine learning based feature extraction in this context exemplifies a progressive analytic strategy to unravel complex biochemical phenomena, offering potential pathways for the refinement of diagnostic tools and deepening the understanding of neurodegenerative diseases from an analytical chemistry perspective.<br>This data was used to train and evaluate the machine learning models as described in Enzlein, et. al., 2024.