Honors Theses
Date of Award
5-2026
Document Type
Undergraduate Thesis
Degree Name
BS
Department
Electrical and Computer Engineering
Faculty Mentor
Mohamed Shaban, Ph.D.
Advisor(s)
Mohamed El-Sharkh, Ph.D., Edmund Spencer, Ph.D.
Abstract
Breast cancer is one of the most common cancer types, and is the second leading cause of death from cancer in women in the United States. Early detection is essential for a better prognosis for the patient, reducing the intensity of the treatment and allowing for a better outcome. To help achieve this early diagnosis, deep learning models can be applied to the different modalities of diagnosis, such as mammograms, ultrasounds, and histopathology images. This project first seeks to develop an improved attention gate for U-Net, an image segmentation model, to help improve the performance without increasing computational complexity on breast ultrasound images. This project also seeks to test various object detection and image classification models on breast histopathology images, and to see if a two-stage object detection to classification pipeline can improve results. The work completed in both parts of the project will help pave the way for future implementations of deep learning, especially in real-time situations on resource-constrained edge devices. Future work for this project will focus on a full end-to-end vision transformer-based system, to compare the performance of transformers to traditional CNN-based models.
Recommended Citation
Potter, Madeline, "Deep Learning Based Detection of breast Cancer in High Resolution Ultrasound and Histopathology Images" (2026). Honors Theses. 144.
https://jagworks.southalabama.edu/honors_college_theses/144
Included in
Bioimaging and Biomedical Optics Commons, Biomedical Commons, Computer Engineering Commons, Health Information Technology Commons, Investigative Techniques Commons, Oncology Commons, Other Biomedical Engineering and Bioengineering Commons, Other Electrical and Computer Engineering Commons, Women's Health Commons