Towards Precise Oncology

Ritika Lama
Dr. Nazmul Shahadat, Faculty Mentor

Breast cancer is identified by the abnormal growth of cells in breast tissue. According to WHO, approximately 99% of breast cancers occur in women, and about 1% occur in men. Symptoms like a breast lump, dimpling, redness, etc., can be seen when cancer is more advanced; those symptoms are mostly not experienced in the early stages. People tend to ignore “minor” symptoms, leading to one of the fatal diseases. Despite medical advancement, there is less chance of survival from cancer in late stages. So, in this project, a computationally efficient deep learning model is proposed using a lightweight MobileNetV3 as a small backbone and E2Net and SCSENet as feature extraction and attention-based mechanisms. The cellular data from Mammogram Mastery is trained, tested, and validated to detect breast cancer, achieving an accuracy of 96%. The model has similar accuracy in classification as other machine learning techniques like Random Forest, J48, SMO, etc.

Keywords: Machine Learning, Mammogram, Cancer Detection, E2Net, SCSENet, MobileNetV3, Medicine

Topic(s):Biology
Computer Science

Presentation Type: Oral Presentation

Session: TBA
Location: TBA
Time: TBA