Enfoque Híbrido para la Clasificación de Muestras Histológicas para la Detección de Tumores Cerebrales: Descomposición lineal y Redes Neuronales

Abstract

Hyperspectral imaging has demonstrated its potential to provide information of the chemical composition of tissue and also of its morphological characteristics. However, discerning the presence of a pathology through this information is not a simple task. Because of this, a hybrid methodology is proposed in this work, which combines the identification of characteristic components present in a hyperspectral image from linear unmixing methods, and the ability to distinguish patterns from a neural network. The results of this research show that the proposed method can distinguish a tumor condition from histological brain samples with an average accuracy of 86%. The study demonstrates the potential of hybrid classification methodologies in the analysis of spectral information for the identification of histological samples affected by tumor tissue.