A Hybrid Deep Learning Model for Breast Cancer Detection and Classification

One of the main areas of study in diagnostic radiology and medical imaging is computer-aided diagnosis (CAD). In reality, a significant number of CAD systems have been used to help doctors identify breast tumours early on mammograms. Medical image analysis algorithms are helpful in providing a better and more accurate comprehension of medical images as well as in boosting the dependability of medical images in the healthcare system because therapy and illness diagnosis are so crucial in medical imaging. Leveraging advanced machine learning techniques has become an invaluable tool for healthcare professionals, enhancing early detection and personalizing treatment plans.Therefore, in this work, we began by mentioning several cutting-edge techniques for detecting breast cancer using a deep learning methodology. The primary drawback of current research is that existing models only concentrate on identifying or detecting benign or malignant tumours, rather than specific types of tumours such adenosis, phyllodes tumour, or lobular carcinoma. We used a number of lightweight deep learning models, like ShuffleNet, to create a flexible model. [11]Additionally, in order to achieve recognition, we create the Resnet50 classical CNN model based on transfer learning.By incorporating transfer learning, the model can effectively use pre-trained networks to enhance its learning capability, potentially yielding better performance than training from scratch.number of lightweight deep learning models, like ShuffleNet, to create a flexible model. [4] A new multi-label breast cancer tissues classification model that combines the benefits of Resnet and the Attention mechanism is also taken into consideration.This innovative approach is designed to focus on specific regions of interest within the images, which can potentially lead to a higher accuracy in detecting subtler signs of various tumour types.