Breast Histology Slides Classification using Fused Lightweight Deep Features

One of the most serious illnesses that affect women is breast cancer (BC), hence early detection and treatment are crucial. The course of treatment for BC is determined by its stage; hence, histology slide-supported BC confirmation is frequently used in medical facilities. The purpose of this study is to provide a method for classifying selected histology slides into the Benign/Malignant class using a binary classifier used with Lightweight Deep Learning Method (LDLM). The suggested approach has several stages: gathering and resizing images; extracting features using a selected LDLM; reducing features with 50% dropout and fusing features to create a new feature vector; and binary classification with 3 -fold cross validation and performance confirmation. Individual deep features (ID) and fused deep features (FD) are used in the suggested experimental work, which was conducted using the pre-trained LDLM for the investigation. The created scheme’s effectiveness is confirmed by binary classification with the SoftMax classifier, and the obtained outcomes are contrasted and validated. The study’s experimental results validate that the suggested methodology yields >89% detection accuracy with the ID and >98% accuracy with the FD.