Proceedings Article | 8 March 2013
KEYWORDS: Raman spectroscopy, Skin, Tissues, Skin cancer, Mahalanobis distance, Tumor growth modeling, Cancer, In vivo imaging, Biological research, Data modeling
This work aimed the development of a discriminating model, using Raman spectroscopy, based on the estimated concentration of biochemical components presented in skin, for in vivo diagnosis. Raman spectra were collected in patients who underwent excision surgery of suspicious lesions at the lesion site and at a normal circumjacent site. It has been estimated the relative amount of selected biochemical compounds presented in skin. The Raman spectra of normal and malignant (basal cell carcinoma - BCC and squamous cell carcinoma - SCC) skin are quite similar, with some spectral differences in the regions of lipids, nucleic acids, and hemoglobin. Some biochemicals showed statistically significant differences among N, BCC and SCC, such as elastin, ceramide, melanin, nucleid acid, actin and phenylalanine. Elastin and ceramide presented significant differences between N and BCC, melanin, DNA and actin presented significant differences between N and BCC and between N and SCC, being melanin and DNA decreased in neoplasias, in contrast with actin, that increased in neoplasias. Concentration of phenylalanine was significantly increased for SCC compared to N and BCC. The relative concentration of melanin, DNA and phenylalanine showed sensitivity, specificity and accuracy of about 81%, 65% and 60%, respectively, using Mahalanobis distance as a discriminator. This model is being incorporated to a Raman system with automated data collection and processing that could be used for a future in vivo, real time discrimination algorithm.
