FTIR spectroscopy imaging coupled with machine learning reveals biochemical changes in the brains of diabetic mice

Diabetic encephalopathy is a progressive complication of type 2 diabetes, yet its region-specific biochemical changes remain unclear. In this study, we applied Fourier Transform Infrared Microspectroscopy (FTIRM) to assess metabolic alterations in the anatomical structures of the brains of db/db mice. Spectral data from five anatomically defined regions, cerebral cortex, hippocampus, thalamus, hypothalamus, and striatum, were analyzed in both 12-week and 21-week wild-type and diabetic groups. Marked changes were observed in the spectral regions of glucose, phospholipids, esterified lipids, amide, and nucleic acids across these brain regions. As diabetes progress, the hippocampus exhibited progressive molecular deterioration, while the hypothalamus showed disruptions followed by partial restoration at 21 weeks. Partial least square discriminant analysis (PLS-DA) models based on second-derivative spectra achieved high classification accuracy (area under the curve (AUC) > 0.99), reliably distinguishing between healthy, early (12 weeks), and late (21 weeks) diabetic stages. These results demonstrate the capacity of FTIRM combined with machine learning to reveal region-specific biochemical signatures of diabetic encephalopathy and offer potential for early diagnosis and staging.