Fluorescence Profiling of Water-Based Breast Tissue Homogenates Combined With Chemometric Analyses for Discrimination of Benign and Malignant Lesions.

Abstract

Breast cancer diagnosis is clinically challenging, particularly in distinguishing benign lesions, ductal carcinoma in situ (DCIS), and invasive ductal carcinoma (IDC). This study evaluates multimodal autofluorescence spectroscopy combined with chemometric analysis for breast tissue classification. A total of 145 ex vivo specimens (56 IDC, 54 DCIS, 35 benign) were analyzed using excitation spectroscopy, synchronous fluorescence spectroscopy, and two-dimensional integrated-emission mapping (2D-IEM). Spectral data were decomposed via parallel factor analysis (PARAFAC) and assessed using PCA with Bonferroni-corrected ANOVA and receiver operating characteristic analysis. Excitation fluorescence demonstrated good discrimination for benign (area under the curve, AUC = 0.85) and IDC (AUC = 0.83) tissues, with moderate DCIS performance (AUC = 0.75). Synchronous modality enhanced differentiation of overlapping fluorophores, revealing tumor-specific metabolic signatures. PARAFAC resolved three biologically relevant components-protein-related, metabolic, and porphyrin-associated-enabling clear tissue group separation. These findings establish multimodal autofluorescence spectroscopy as a robust, label-free approach with strong potential for intraoperative margin assessment and diagnostic support.