The clinical differential diagnosis of skin tumors is an often-challenging task, to which the probing of skin with mid- and near-infrared (IR) light may be contributory. The development of objective methods for the analysis of IR spectra remains a major hurdle to developing clinically useful applications. The authors highlight different processing methods for IR spectra from skin biopsies and in-vivo skin tumors. Spectroscopic maps of biopsies of basal cell, squamous cell and melanocytic neoplasms were objectively grouped into distinct clusters that corresponded with tumor, epidermis, dermis, follicle and fat. Normal and abnormal skin components were located within maps using a search engine based upon linear discriminant analysis (LDA). In all instances, areas of tumor were distinct from normal tissue in biopsies. In-vivo, near-IR spectroscopy and LDA allowed discrimination between benign and malignant skin lesions with a high degree of accuracy. We conclude that IR spectroscopy has significant diagnostic promise in the skin cancer arena. The analytical methods described can now be used to create a powerful classification scheme in which to detect skin tumor cells within biopsied and living skin.
|