According to the National Brain Tumor Society, around 700,000 people are living with a primary brain tumor in the United States. A newly developed artificial intelligence could be a future tool for brain tumor diagnosis and treatment.
Neurosurgeons and engineers from the University of Michigan Medicine have developed an AI-based diagnostic screening system called DeepGlioma. The system uses rapid imaging to analyze tumor specimens removed from an ongoing operation and can successfully detect genetic mutations in brain tumors in less than 90 seconds. Todd Hollon, M.D., Lead Author and Creator of DeepGlioma, highlights the new discovery in a University of Michigan release. “This AI-based tool has the potential to improve the access and speed of diagnosis and care of patients with deadly brain tumors,” Hollon says. “DeepGlioma creates an avenue for accurate and more timely identification that would give providers a better chance to define treatments and predict patient prognosis.”
A glioma is a common form of tumor located in the brain that arises from glial cells. John Hopkin’s Medicine says nearly 33% of all brain tumors are gliomas. While there are many different glioma forms, the new AI study focuses specifically on diffuse gliomas.
Diffuse gliomas earn their name from their ability to surround the brain tissue by utilizing diffuse infiltration. The most occurring forms of diffuse gliomas include astrocytomas, oligodendrogliomas, and mixed oligoastrocytomas.
New AI prediction model
The new AI study, published in Nature Medicine on 23 March 2023, included 153 people with diffuse glioma. The University of Michigan served as the home for model training, while New York University, the University of California San Francisco, the Medical University of Vienna, and University Hospital Cologne served as patient enrollment centers.
Researchers combined stimulated Raman histology and deep learning-based image classification to predict molecular genetic features used by the WHO (World Health Organization) to define diffuse glioma. Stimulated Raman histology was developed by the University of Michigan researchers in 2019 to image brain tumor tissue instantly.
In their results, researchers were able to achieve average molecular genetic classification accuracy of 93.2%. Investigators identified the correct diffuse glioma molecular subgroup with 91.2% accuracy within two minutes while in the operating room. It is vital to have the correct molecular classification of glioma to diagnose and treat the tumor. The specific form of diffuse glioma may determine the lifespan of a patient. Individuals with a diffuse astrocytoma can have five additional years of life versus another diffuse glioma subtype.
Hollon says DeepGlioma gives medical care providers the opportunity to be more accurate in identifying the type of brain tumor. Currently, few options are available for determining the type of brain tumor.
Brain Tumors in the U.S.
For Americans in 2023, the American Society of Clinical Oncology (ASCO) estimates 24,810 adults (14,280 males and 10,530 females) will be diagnosed with primary cancerous tumors in the brain or spinal cord. Children can also be affected by brain tumors, the ASCO says 5,230 youngsters under 20 years will be diagnosed with a central nervous system tumor in 2023.
Despite the likelihood of an individual developing a brain or spinal cord in their lifetime at 1%, brain and nervous system cancers are the tenth leading cause of death for both men and women. In 2023, an estimated 18,990 Americans (11,020 males and 7,970 females) will die from brain and spinal cord tumors. ASCO notes brain tumors are more common in children and older adults, however, individuals can be diagnosed at any age. Family history has a role to play in brain tumors, with 5% due to genetic factors.
Senior Author of the study and associate professor of neurosurgery and pathology at NYU, Daniel Orringer, M.D., tells Michigan Medicine little has been done to advance deadly brain tumor treatment. “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,” Orringer says. “Rapid methods for molecular classification hold great promise for rethinking clinical trial design and bringing new therapies to patients.”
REFERENCE: Health News; 27 MAR 2023; Erick Mitchell