Spielman grant research utilizes machine-learning in place of biopsies
The intersection of integrated systems engineering and radiation oncology may lead to breakthrough medical care that is less invasive for post-op breast cancer patients.
That’s the hope behind “Predicting Pathologic Complete Response Using Machine Learning and Magnetic Resonance Imaging (MRI) Data in Breast Cancer Patients,” which was awarded a two-year, $50,000 grant through the Stefanie Spielman Pilot Research Grant Program.
ISE Assistant Professor Samantha Krening and Dr. Sachin Jhawar, an associate professor in the Department of Radiation Oncology at the Stefanie Spielman Comprehensive Breast Center, are serving as co-principal investigators on the project. Dr. Jeffrey Hawley, clinical associate professor, and Dr. Clayton Taylor, associate professor, both in the Department of Radiology at The Ohio State University Comprehensive Cancer Center, will serve as co-investigators.
Currently, breast cancer patients who have had surgery undergo a biopsy to determine if any cancer cells remain. “What we want to do instead is give them an MRI,” Krening says.
She is working to design an algorithm that would go beyond the current findings of an MRI and pinpoint where the cancer is as opposed to a biopsy “which would just denote that cancer cells are still present,” she says.
Through multilayer neural networks, or deep learning, machines now have the ability to identify significant features as the machine processes images, according to Krening. “How loud is this signal in the data?” she asks. “We can tailor the algorithm. Maybe in the beginning, it won’t have the exact location, but I think it will.”
The method could have numerous benefits for the patient over biopsies. In addition to lower cost and being less time-consuming, it also could relieve a lot of stress, Krening says.
“One thing it could help us do is escalate or de-escalate treatment,” she says. “If something isn’t working, we can respond earlier while in treatment. Time matters.”
Krening is relying on Dr. Jhawar to provide a large data set of breast cancer imaging utilizing five years of retrospective data from The James’ Institutional Review Board. Dr. Hawley and Dr. Taylor are demonstrating how they analyze MRIs, which will inform Krening’s algorithms for machine-learning and “how to structure and improve accuracy,” she says.
In addition to a graduate student working on the project, Krening’s ISE undergraduate students also will be studying the data.
“There are a lot of benefits,” she says. They get experience with very different data – medical data that is not easy to work with. They will have a lot of expertise when they graduate and will have already cut their teeth on some of these very difficult projects.”
In addition, she says, it allows students to have contact with other researchers outside the ISE Department. Krening met Dr. Jhawar through one of her students who worked as a radiation physicist. Since then, Krening has been teaming up with him on other projects, including head and neck cancer studies.
“It’s so nice to have access to all these amazing doctors to collaborate with and do meaningful research,” she says.
She anticipates that the work they are doing with the MRI results could be expanded to other cancer patients and may result in follow-on funding through proposals with the National Institutes of Health and the National Science Foundation.
Story by Nancy Richison
