Using Machine Learning for Assistance in Early Breast Cancer Detection
Lupe Amigon, Isaac Walker, Manuel Duran
Department of Computer Science
Faculty Supervisor: Sara El Alaoui
Breast cancer affects approximately 1 in 8 women in the US and mammograms, low dose x-rays of the breast, are a crucial tool used for breast cancer prevention as it helps detect the disease at its early stages. For our model training we used the CBIS-DDSM dataset, which contains ten thousand categorically sorted mammograms. We will create machine learning models that will be trained and validated on this image bank, identifying progression and presence of cancer, and classifying said images in three categories: benign, normal, or malignant. We aim to create a model that can accurately differentiate and classify these categories. Applications of models like these could assist future radiologists and doctors diagnose breast cancer, reduce misdiagnosis, and give patients ease of mind by reducing turnaround time.