Ph.D. student helps improve efficacy of mammograms through research program
Folami Alamudun spent his summer in the Research Alliance Math and Science Program at ORNL, analyzing the visual scanning pattern radiologists use to detect cancer in mammograms. His goal is to help increase the efficiency and efficacy of these screenings through improved detection tools.
Folami Alamudun, fourth-year doctoral student at Texas A&M University Sketch Recognition Laboratory (SRL) spent 10 weeks at the national lab housing some of the world’s most powerful processors. The decision to accept the opportunity, however, did not come easily.
“My wife and I made a very difficult but necessary decision to spend the summer apart so I could focus on making much needed progress here at Oak Ridge National Laboratory,” said Alamudun, a participant in the U.S. Department of Energy’s Research Alliance Math and Science (RAMS) Program.
The RAMS Program invites undergraduate and graduate students to intern at ORNL for 10-12 weeks. The program is designed to provide a collaborative research experience that promotes science, technology, engineering and math and encourages the creation of a diverse national laboratory workforce.
Alamudun continued, “The decision was made much more difficult because we anticipated the arrival of our twin children during this time.”
In the end, everything turned out well: Alamudun was able to travel back to Texas to visit the newborns, and when he returned to ORNL he contributed research that could decrease false negatives on mammogram cancer screenings. It was research that earned him not only a Best Poster award during the end-of-summer presentation session but also a spot for presenting at the SPIE Medical Imaging Conference in Florida in February 2015.
Typically, radiologists will view a mammogram on four panels. Abnormal screenings will show a heavier concentration of white in a certain area of the breast, which is often easy to spot. Other times, however, being tuned into a potential problematic area requires an advanced level of perception and comprehension. Over time, radiologists develop search patterns that help them identify these problem spots before they mature into full-blown cancer.
Alamudun’s objective was to understand and model these patterns. Specifically, he analyzed the complexity of ocular fixations collected using a head-mounted eye-tracking device. The data collection was done at the radiologists’ regular work settings. These gaze data were then analyzed to identify common properties across radiologists that could be used to develop personalized visual aids for computer-aided diagnostic tools and training purposes.
Alongside his mentor Georgia Tourassi, Ph.D., in ORNL’s Biomedical Sciences and Engineering Center, Alamudun developed data analysis skills that contribute to his long-run professional goals of conducting computational analytics research at a national lab. Â He imagines using computers to improve human efficiency, energy acquisition, utilization and conservation, and environmental sustainability.
The RAMS program was a stepping stone to that vision, an opportunity to form relationships with world-renowned scientists while utilizing cutting-edge technology.
“My favorite part of the program was meeting with the people here,” said Alamudun. “ORNL has somehow created a mix of very diverse and approachable people. This makes it a great place to perform research. I am almost overwhelmed with the possibility to learn about and participate in areas of research outside of my experience and understanding.”
Additionally, his experience with his mentor, Tourassi, encouraged him to consider participating in that kind of capacity one day, as well.
“Let me state emphatically, that without the support and guidance of my mentor, this experience would have been completely impossible. She instills confidence in the most unassuming way,” he said. “I have observed her very keenly during my time here, and I hope to emulate and be of similar value to younger scientists in the future.”