Deciphering cross-scale spatiotemporal mapping and metabolic signatures of aflatoxins accumulation in Aspergillus-infected peanuts.

Aflatoxin contamination in peanuts poses carcinogenic risks, yet conventional detection methods provide limited rapid, non-destructive assessment and insight into the dynamic spatiotemporal progression and associated metabolic changes. This study integrated near-infrared hyperspectral imaging (NIR-HSI), mass spectrometry imaging (MSI), and ultra-high-performance liquid chromatography-tandem mass spectrometry (uHPLC-MS/MS) with machine learning (ML) to elucidate the cross-scale spatiotemporal mapping and metabolic signatures associated with aflatoxin accumulation in Aspergillus-infected peanuts. The NIR-HSI-based predictive model achieved 100% classification accuracy in discriminating control, toxigenic, and nontoxigenic strains. Quantitative models achieved Rcv/Rpre values of 0.94/0.84 and 0.87/0.90 for AFB1 and AFB2, respectively. MSI combined with ML analysis resolved the microscale spatiotemporal metabolic dynamics of aflatoxins and their metabolites across various infection stages. The early stage (0-24 h) was characterized by lipid oxidation and synthesis of flavonoid antimicrobials; the mid-stage (24-48 h) was accompanied by fungal carbohydrate plundering; and the later stage (48-72 h) exhibited systemic metabolic disorder. NIR-HSI and MSI collectively demonstrated that Aspergillus flavus infection typically initiates spatially on the surface near the hypocotyl and progressively advances inward, consistent with the aflatoxins' accumulation dynamics. In conclusion, this work established a cross-scale spatiotemporal framework, providing a foundation for early warning and prevention of aflatoxin contamination.