Predictive Analysis for Early Detection of Breast Cancer Through Artificial Intelligence Algorithms

Breast cancer is the most common type of cancer in women worldwide. In Italy, there were 59,700 new diagnoses in 2023. Early diagnosis is crucial for fighting cancer and reducing the mortality rate. Ultrasound imaging is a common technique used in the early diagnosis of breast cancer. The paper provides the radiologist with the mask of the tumor mass mask along with its classification to make the diagnosis more effective. The proposal examines different deep learning models for image classification and uses the selected best model to perform segmentation in order to create an image mask to identify the tumor mass. The BUSI dataset is one of the few publicly available datasets. Experimental studies have shown that classification models can successfully classify breast ultrasound images that segmentation models can produce a good mask. The best-performing model outperforms other models with an accuracy of over 90%, a precision of 92%, a recall of 90%, and an F1-score of 90%. This study shows that deep learning architectures are effective in classifying and segmenting ultrasound breast images and could be used in clinical experiments in the near future.