LSU’s Seungwon Yang, PhD Leads Research Group Using AI to Detect Breast Cancer

Associate Professor & Russell B. Long Professor Seungwon Yang holds a joint position with the School of Library & Information Science (SLIS) and the Center for Computation & Technology (CCT). During the past year, Yang collaborated with multiple universities to leverage artificial intelligence (AI) in the identification of breast cancer within mammogram images. Yang’s use of AI grew from discussions with colleagues at the Deep Learning Summit in Boston. As Yang discovered, breast cancer diagnosis from mammogram images remains shockingly low and inconsistent among radiologists. The low diagnosis rates frequently require patient reexamination with repeated radiation exposure and missed cancer cells.

Returning from the Deep Learning Summit, Yang assembled a team of doctoral students at CCT to focus on a deep learning algorithm that would find potentially cancerous regions and diagnose these regions for cancer cells. The group faced early challenges with data sharing between universities, but found success accessing an existing dataset of mammogram images from the University of South Florida, allowing them to continue their research.

               The use of this AI deep learning algorithm for cancer detection is one of the first of its kind. It can automatically detect and diagnose areas based on a single mammogram image. If the newly developed algorithm is successful, a patient will be able to upload one image to the framework and the AI will detect suspicious regions of the mammogram and diagnose the region if something suspicious is found. While the initial results are promising, Yang and his research team are hoping to achieve higher performance as they gather more real-life mammogram data.

Through its initial development and testing, the AI algorithm performed very well, achieving a specificity of 93%, sensitivity of 94%, accuracy of 93.5%, precision of 93.39%, recall of 93%, and area under the curve (AUC) of 92.315%. An AUC of 100% would represent a perfect system, but an AUC in the range of 60-90% is considered good with anything over 90% demonstrating high performance. Yang and his team hope they will be able to assist with the advancement of accurate, reliable, and automated breast cancer diagnoses as their research move forward.

About the LSU School of Library & Information Science


The LSU School of Library & Information Science (SLIS) provides a 100% online prestigious education in library and information science. It is the home of the Master of Library and Information Science, which is the only program accredited by the American Library Association in the state of Louisiana. SLIS also offers a dual degree with the Department of History, an undergraduate minor, and three graduate certificate options. SLIS is a member of the iSchools, a group of Information Schools dedicated to advancing the information field. SLIS is part of the LSU College of Human Sciences & Education.    


For more information, visit the School of Library & Information Science.


About the LSU College of Human Sciences & Education

The College of Human Sciences & Education (CHSE) is a nationally accredited division of Louisiana State University. The college is comprised of the School of Education, the School of Kinesiology, the School of Leadership & Human Resource Development, the School of Library & Information Science, the School of Social Work, and the University Laboratory School. These combined schools offer 8 undergraduate degree programs, 18 graduate programs, and 7 online graduate degree and/or certificate programs, enrolling more than 1,900 undergraduate and 1,120 graduate students. The College is committed to achieving the highest standards in teaching, research, and service and is committed to improving quality of life across the lifespan.  


For more information, visit the College of Human Sciences & Education