Enhanced Bag of Features Using AlexNet and Improved Biogeography-Based Optimization for Histopathological Image Analysis

Bag of features is an efficacious method for image classification. However, its applicability on histopathological images is still an open ended research problem. In this paper, a novel bag of features based histopathological image classification method is presented. The proposed method involves three steps: (i) Feature extraction using AlexNet, (ii) Optimal visual vocabulary generation using improved biogeography-based optimization, and (iii) Classification using support vector machine. The experimental evaluation is conducted on the standard histopathological image dataset namely; Animal Diagnostics Lab (ADL) dataset having images of three organs as kidney, lung, and spleen. Each organ has inflamed and healthy tissue images. The performance of proposed method is compared with five state-of-the-art histopathological image classification methods in term of precision, recall, F1-score, and overall average accuracy. Simulation results show that the proposed method outperforms other considered state-of-the-art methods.