Application of Evolutionary Algorithms on Unsupervised Segmentation of Lymphoma Histological Images

Abstract

Histological images analysis is widely used to carry out diagnoses of different types of cancer. Digital image processing methods can be used for this purpose, leading to more objective diagnoses. Segmentation techniques are applied to identify cellular structures indicative of diseases. In addition, the extracted features from these specific regions can aid pathologists in diagnoses decision using classification techniques. In this paper, we present an evaluation of evolutionary algorithms applied to lymphoma images for segmentation of their neoplastic cellular nuclei. In a second stage, we investigated the performance of the segmented images in the classification step. Initially, the R channel from RGB color model was processed with histogram equalization and Gaussian filter. In the segmentation step, optimization methods were analyzed in combination with the fuzzy 3-partition technique. Then, we also applied the valley-emphasis method and morphological operations to remove false positive regions in the post-processing step. Intensity and texture features were extracted and classified by the support vector machine method for diagnoses of 62 and 99 images of follicular lymphoma and mantle cell lymphoma, respectively. The results were evaluated through qualitative and quantitative analyses and the differential evolution method has reached the best results in the segmentation step. This technique allowed a relevant performance on the classification task with a mean value of accuracy of 99.38%.