Benign-Malignant Breast Histology Classification using MobileNet Variants: An Analysis

Department of Computer Science and Engineering, Saveetha School of Engineering, Saveetha Institute of Medical and Technical Sciences (SIMATS), Chennai 602105, TN, India

^* Corresponding author: This email address is being protected from spambots. You need JavaScript enabled to view it.

Abstract

Automatic analysis of the medical data is one of the common practices followed to detect diseases with better accuracy. Deep Learning (DL) tool-based medical image examination is one of the approved clinical practices, and the outcome of this process supports the treatment planning and execution. This work proposes a DL tool based on the ConvNeXt (CN) scheme to classify the chosen Breast Histology Images (BHI) into benign and malignant classes. The various phases of the proposed DL-tool include: image collection from the database and resizing it to 224x224x3 pixels, feature extraction using the chosen CN-model, feature reduction using 50% dropout, and serial features fusion to get fused- features-vector (FFV), and binary classification with 5-fold cross-validation. The merit of the developed scheme is confirmed using the classification executed with the chosen CN-feature and the FFV. The outcome of this study confirms that the FFV-based classification provides a detection result upto 99% when the SoftMax-based classification is executed. This confirms that the proposed DL-tool provides a better result on the chosen image database.