Breast Cancer classification via Deep Learning approaches

Breast cancer is the most common type of cancer in women worldwide. In 2023, there were 2.296.840 (23,8% of all women with cancer) new diagnoses. Early diagnosis is a key factor in reducing the mortality rate of breast cancer. One of the screening methods used to prevent breast cancer is breast ultrasound. In this paper, a new model is proposed that starts from a resnet101 and increases the classification capacity of a normal resnet101. Experimental studies show how deep learning models can successfully classify breast ultrasound images. The proposed model achieves 91% accuracy with convergence in less than 30 epochs. This study shows that deep learning models are effective in classifying ultrasound images and could be used by a radiologist to increase the accuracy of diagnoses.