A nuclei segmentation method based on butterfly algorithm for h&e stained images

Abstract

Nuclei segmentation in H&E strained images plays a vital role in diagnosis of various diseases. Huge research is being carried out by various researchers in developing computerized methods for automatic segmentation of nuclei. These computerized methods played vital role in minimizing human intervention in diagnosis of various diseases. In this paper, we have proposed a new nuclei segmentation method which uses Butterfly Algorithm for avoiding local optima for Histopathological images. The algorithm is based on food foraging strategy of butterflies, as they use their sense of sight, taste, smell, touch and hearing to determine the position of food and mating partner. Histopathological images data set of TNBC patients has been taken. The performance measure of the proposed method is evaluated bases on Accuracy, F1 score and Aggregated Jaccard Index.