A Novel Approach for Detecting Breast Cancer Cells and Comparison using DeepLearning Techniques

Breast cancer is a major worldwide health issue, highlighting the need for precise and effective diagnostic instruments. The research thoroughly examines the use of deep learning methods for detecting breast cancer by using three advanced CNN models: InceptionV3, VGG16, and MobileNetV3Large. The goal is to evaluate the performance and effectiveness of these algorithms in detecting breast cancer characteristics in medical imaging data. This study used a dataset including a wide variety of mammography pictures, including both benign and malignant instances. The deep learning models are pre-trained on extensive datasets and then adjusted on the particular breast cancer dataset to improve their capacity to detect subtle patterns that suggest malignancy. The research assesses the accuracy, sensitivity, specificity, and overall performance of the models in differentiating between benign and malignant lesions. The results show positive outcomes for all three models, each displaying a different level of skill in detecting breast cancer. The InceptionV3 model, with its intricate design using inception modules, has a strong capacity to record intricate spatial hierarchies. VGG16 is recognized for its straightforward design and many layers, demonstrating strong performance in extracting features. MobileNetV3Large is designed for mobile and edge computing applications, showcasing efficiency in resource use while maintaining accuracy. The comparison study offers a detailed examination of the strengths and weaknesses of each model, assisting in choosing a suitable deep learning architecture for breast cancer detection tasks. The research also explores possible ways to improve the model and its applicability to real-world clinical environments. This study provides vital insights to the expanding area of medical image analysis, facilitating the creation of more precise and dependable instruments for early breast cancer diagnosis using deep learning methods.