Researchers have developed an AI-assisted method to screen for genetic mutations in cancerous brain tumors within 90 seconds. This discovery could revolutionize how gliomas, a type of tumor, are diagnosed and treated.
Michigan Medicine neurosurgeons, engineers, and investigators from New York University and the University of California, San Francisco, created DeepGlioma — an AI-based diagnostic system. It utilizes rapid imaging to identify genetic mutations quickly in specimens taken during surgery.
Using the recently developed system, researchers from a study involving more than 150 patients with diffuse gliomas demonstrated an average accuracy above 90% in identifying mutations determined by the World Health Organization’s molecular subgroups. These findings have been published in Nature Medicine, highlighting this deadly primary brain tumor.
Dr. Todd Hollon, a neurosurgeon and assistant professor of neurosurgery at U-M Medical School, created DeepGlioma and is the lead author of it. He’s part of the University of Michigan Health care system.
Todd Hollon says:
“This AI-based tool has the potential to improve the access and speed of diagnosis and care of patients with deadly brain tumors,”
Molecular classification of gliomas is becoming ever more important in planning the course of treatment, as the genetic makeup of each brain tumor patient determines what risks and benefits should be considered before surgery.
Complete tumor removal can lead to an average of five extra years for patients with astrocytomas, a specific type of diffuse glioma. Therefore, astrocytoma patients are uniquely advantaged compared to other diffuse glioma subtypes.
Molecular testing for diffuse glioma is not accessible or consistent among brain tumor treatment centers. Additionally, according to Hollon, when available, the duration of getting results can be lengthy — ranging from days to weeks. As a result, making use of such testing may be difficult for many.
Hollon says:
“Barriers to molecular diagnosis can result in suboptimal care for patients with brain tumors, complicating surgical decision-making and selection of chemoradiation regimens,”
In 2019, U-M developed a revolutionary system called DeepGlioma that uses deep neural networks and an optical imaging method called stimulated Raman histology to differentiate diffuse gliomas during surgery. Before this, surgeons did not have the capability of doing this.
Hollon went on to say:
“DeepGlioma creates an avenue for accurate and more timely identification that would give providers a better chance to define treatments and predict patient prognosis,”
Despite optimal standard-of-care treatment, the prognosis for those with malignant diffuse gliomas remains poor, with a median survival time of 18 months. Treatment options are, therefore, limited.
Researchers are hopeful that DeepGlioma can encourage early participation in clinical trials for glioma, although these trials often have limited molecular subgroup enrolment. This is especially crucial as fewer than 10% of patients with glioma participate in such trials despite needing medication development to treat the tumor.
Daniel Orringer, M.D., an associate professor of neurosurgery and pathology at NYU Grossman School of Medicine, is the senior author who pioneered stimulated Raman histology. He developed this technique.
Daniel Orringer says:
“Progress in the treatment of the most deadly brain tumors has been limited in the past decades- in part because it has been hard to identify the patients who would benefit most from targeted therapies,”
“Rapid methods for molecular classification hold great promise for rethinking clinical trial design and bringing new therapies to patients.”
The use of AI technology to predict the genetics of cancerous brain tumors in under 90 seconds represents a major step forward in the fight against cancer. As our understanding of the human genome grows, AI-powered tools will undoubtedly play an increasingly important role in developing new treatments and cures for various diseases and medical conditions.
Source: medicalxpress.com
