University of Virginia School of Medicine scientists are tapping the power of artificial intelligence (AI) to enhance and accelerate treatment for glioblastoma, the deadliest brain cancer.
UVA researcher Dr. Bijoy Kundu and colleagues are developing an AI imaging approach to distinguish between tumor progression and brain changes caused by tumor treatment. The process can take months to make that distinction, leaving doctors uncertain if the tumor is growing and stalling important care decisions.
Kundu’s AI approach is already outperforming the standard clinical option in initial testing. Put to the test in 26 glioblastoma patients immediately after treatment, the artificial intelligence was able to make the correct distinction 74 percent of the time.
“The goal of the project is to train on additional patient data and increase this accuracy by greater than 80 percent for clinical use,” said Kundu, of UVA Cancer Center, UVA Health’s Department of Radiology and Medical Imaging and UVA’s Department of Biomedical Engineering.
The results could have real benefits for patients.
“Early distinction would enable earlier treatment modifications for tumor recurrence in brain cancer patients,” said Dr. David Schiff, part of UVA’s Departments of Neurology, Neurosurgery and Medicine. Schiff is also co-director of UVA Health’s Neuro-Oncology Center.
Glioblastoma makes up more than half of all primary brain tumors and is a highly aggressive and fast-growing cancer with typical survival from diagnosis only 15 months. The short survival time makes it extremely important that doctors act quickly. Doing so can extend survival time and improve patients’ quality of life.
Now, however, doctors must wait three to four months after treatment to assess the progression of the tumor by using magnetic resonance imaging (MRI) or, in some cases, brain surgery.
Kundu’s approach combines MRI with another form of imaging: dynamic PET (positron emission tomography), which produces sophisticated, multidimensional views inside the brain that artificial intelligence can analyze – all without the need to cut inside the skull.
Kundu and his collaborators have received $90,000 from UVA’s Ivy Biomedical Innovation Fund to advance and refine their approach. They will use the funding to improve the accuracy of their deep-learning algorithms, essentially teaching the AI to better distinguish between signs of tumor progression and the effects of chemotherapy and radiation therapy.
They hope their work will ultimately help doctors acquire the information they need sooner, improving care for patients with glioblastoma.
“We hope this work helps patients and families get answers faster. If our AI can give doctors more confidence earlier, it could mean quicker treatment decisions and better outcomes. Our goal is to give doctors better tools, so they can focus less on guesswork and more on care. We’re still in the early stages, but even now, our approach is showing real promise. We’re working toward a future where patients get clarity sooner, and where that clarity helps save lives,” Kundu said.
Finding new ways to improve patient care is a core mission of both UVA Cancer Center and UVA’s Paul and Diane Manning Institute of Biotechnology. UVA Cancer Center is one of only 57 cancer centers in the country designated “comprehensive” by the National Cancer Institute for their exceptional patient care and cutting-edge cancer research.
The Manning Institute, meanwhile, has been launched to accelerate the development of new treatment and cures for the most challenging diseases. The new treatment development will be complemented by a statewide clinical trials network that expands access to potential new treatments as they are developed and tested.
UVA’s Department of Biomedical Engineering is a joint program of the School of Medicine and the School of Engineering and Applied Science.
Share
Rebecca Barnabi
