This work aims to investigate and differentiate skin cancers as Basal Cell Carcinoma (BCC) and Squamous Cell Carcinoma (SCC) using a combination of Fourier Transform Infrared (FTIR) spectroscopy, Mueller Matrix Polarimetry (MMP), and Support Vector Machine (SVM) classification. Ex vivo characterization was performed on SCC (malignant) and BCC (malignant and more aggressive) tissue samples collected from 100 different patients. Each sample was analyzed using microscopy for morphological insights, FTIR spectroscopy for biochemical composition across the wavenumber range of 400–4000 cm⁻1, and MMP in the visible range (400–800 nm) for polarimetric behavior. A total of 13 polarimetric parameters were extracted and statistically analyzed. Furthermore, a Machine learning SVM model was trained using 1150 observations and 34 predictors to differentiate between the tissue types. FTIR analysis revealed significant spectral differences between SCC and BCC. Notably, SCC exhibited peaks related to Aldehyde (lipids), Amide I (proteins), Z-DNA (nucleic acids), and Glycosylation, whereas BCC showed prominent signatures of Amide B (proteins), CH₂ stretching (lipids), Aldehyde (lipids), Amide I, B-DNA (nucleic acids), and Glycosylation. Polarimetric analysis through MMP demonstrated statistically significant (p < 0.05) differences in all parameters except latitude, orientation, and ellipticity, which were higher in BCC samples. The SVM classifier achieved 90% accuracy, sensitivity, and specificity, demonstrating strong potential for automated cancer classification. The integrated approach of FTIR spectroscopy and polarimetric analysis, coupled with machine learning via SVM, provides a powerful diagnostic tool for distinguishing between SCC and BCC. This novel methodology supports the potential for developing automated, non-invasive, and accurate systems for early cancer diagnosis.
Siddique et al. (Thu,) studied this question.