Discrimination between basal cell carcinoma and hair follicles in skin tissue sections by Raman micro-spectroscopy

Skin cancer is the most common human malignancy and basal cell carcinoma (BCC) represents approximately 80% of the non-melanoma cases. Current methods of treatment require histopathological evaluation of the tissues by qualified personnel. However, this method is subjective and in some cases BCC can be confused with other structures in healthy skin, including hair follicles. In this preliminary study, we investigated the potential of Raman micro-spectroscopy (RMS) to discriminate between hair follicles and BCC in skin tissue sections excised during Mohs micrographic surgery (MMS). Imaging and diagnosis of skin sections was automatically generated using ‘a priori’-built spectral model based on LDA. This model had 90 ± 9% sensitivity and 85 ± 9% specificity for discrimination of BCC from dermis and epidermis. The model used selected Raman bands corresponding to the largest spectral differences between the Raman spectra of BCC and the normal skin regions, associated mainly with nucleic acids and collagen type I. Raman spectra corresponding to the epidermis regions of the hair follicles were found to be closer to those of healthy epidermis rather than BCC. Comparison between Raman spectral images and the gold standard haematoxylin and eosin (H&E) histopathology diagnosis showed good agreement. Some hair follicle regions were misclassified as BCC; regions corresponded mainly to the outermost layer of hair follicle (basal cells) which are expected to have higher nucleic acid concentration. This preliminary study shows the ability of RMS to distinguish between BCC and other tissue structures associated to healthy skin which can be confused with BCC due to their similar morphology.