Automated Breast Cancer Classification based on Modified Deep learning Convolutional Neural Network following Dual Segmentation

As soon as possible, breast cancer must be diagnosed and treated. Deep learning-based breast cancer classification and segmentation approaches are introduced in this research. A novel computer-aided detection method is described for the classification of normal and malignant mass tumors. This system employs two types of segmentation. The first approach relies on manually determining the ROI, whereas the second makes use of thresholds and a region-based strategy. An AlexNet DCNN framework is used to extract features and categories two kinds of data. Support vector machine classifiers are plugged into the final layer for better accuracy. A high accuracy rate is achieved through training on a vast amount of data. Despite this, due to patient capacity limitations, biomedical databases contain a relatively smaller sample. Thus, image enhancement may be used to increase the amount and quality of input data. Data augmentation may be done in a variety of ways, but rotation is the one used here. Analyzing both types of sample segmentation yielded similar results, the maximum area under the curve (AUC) was 0.88 (88 percent) and the DCNN's accuracy is raised to 88.6 percent. As a result, the SVM accuracy increases to 94.2 percent, with an AUC of 0.94. (94 percent). When compared to earlier work under the identical conditions, this is the greatest AUC value.