Paraconsistent analysis network applied in the treatment of Raman spectroscopy data to support medical diagnosis of skin cancer.

Paraconsistent logic (PL) is a type of non-classical logic that accepts contradiction as a fundamental concept and has produced valuable results in the analysis of uncertainties. In this work, algorithms based on a type of PL-paraconsistent annotated logic of two values (PAL2v)-are interconnected into a network of paraconsistent analysis (PANnet). PANnet was applied to a dataset comprising 146 Raman spectra of skin tissue biopsy fragments of which 30 spectra were determined to represent normal skin tissue (N), 96 were determined to represent tissue with basal cell carcinoma, and 19 were determined to be tissue with melanoma (MEL). In this database, paraconsistent analysis was able to correctly discriminate 136 out of a total of 145 fragments, obtaining a 93.793 % correct diagnostic accuracy. The application of PAL2v in the analysis of Raman spectroscopy signals produces better discrimination of cells than conventional statistical processes and presents a good graphical overview through its associated lattice structure. The technique of PAL2v-based data processing can be fundamental in the development of a computational tool dedicated to support the diagnosis of skin cancer using Raman spectroscopy.