Deep Learning Model-Based Region Measurement Analysis for Breast Cancer Classification

The mortality of Breast cancer patients is issued based on lack of identification and treatment, and mammography is a useful tool for early screening. Deep learning-based computer-aided diagnosis (CAD) of mammography can help radiologists make more accurate and objective decisions. However, a lot of the current approaches rely on datasets that have been manually annotated for segmentation. Furthermore, several techniques are less interested in using ROI for image fusion models because of the high image sizes and modest lesion proportions. These flaws raise the application of the model's labour, cost, and computational overhead. Consequently, a network with regional scoring based on the deep location network (DLN) model is suggested. Utilizing pooling modules, DLN is considered a single feature extractor end-to-end mammography image classification approach that locates lesion sites without the need for training. Different deep learning models have been considered to test our proposed model. The experiments are demonstrated on publicly available IN breast and CBIS-DDSM datasets, and the proposed model fared well when compared to earlier cutting-edge techniques for classifying mammogram images.