Intelligent Data Classification and Computer Vision model-based Breast Cancer Image Processing Using Digital Mammography

Breast cancer poses a global health challenge for women, making early detection and precise classification crucial for successful treatment. The study presents Intelligent Data Classification and Computer Vision model-based Breast Cancer Image Processing Using the Digital Mammography (IDCCBC) Model, a deep learning that enhances the precision of breast cancer detection. ANN, DBN, and CNN-based classification utilize computer vision for data processing and incorporate a unique feature selection technique known as entropy-controlled firefly-based selection. The MIAS dataset consists of 322 annotated mammogram images that are categorized as normal or abnormal. It recognizes various kinds of abnormalities, including calcifications and different types of masses. The CNN model improves tumor detection by a unique method for resizing bounding boxes that incorporate healthy tissue. Two CNN models handle varied tumor sizes with input dimensions of 256×256 and 128×128 pixels. ResNet, EfficientNet, and MaxViT (Vision Transformer-based) enhance performance by addressing challenges such as vanishing gradients and boosting efficiency. The Entropy-Controlled Firefly Algorithm (ECFA) enhances feature selection, concentrating on regions with elevated tumor intensity to improve accuracy and minimize misclassifications.