AI Predicts After-Effects of Brain Tumor Surgery from MRI Scans
Posted on 29 Jan 2025
The removal of a malignant brain tumor, or glioma, can significantly extend a patient’s life, depending on the type of glioma. However, the true impact of surgery on complex cognitive functions remains unclear, and it may mark the beginning of a challenging recovery process. Many patients experience cognitive difficulties, such as issues with concentration and performing complex tasks, after their brain tumor is removed. These challenges can greatly affect their daily lives, severely diminishing their quality of life. While neurological issues like partial paralysis and vision loss are well understood, the effects of surgery on more intricate cognitive functions are less clear, making it difficult to predict how individual patients will be impacted. Now, researchers have developed an artificial intelligence (AI) model that can predict the effects of surgery on cognitive tasks by using neural connection data extracted from a patient’s pre-surgery MRI scans.
The AI model, developed at Eindhoven University of Technology (TU/e, Eindhoven, Netherlands), can assist in predicting how a patient with a malignant brain tumor will perform cognitive tasks post-surgery. The brain’s function relies heavily on neurons that form long-distance bundles, known as white matter, which physically connect different regions of the brain. The researchers used detailed structural information from the white-matter connections visible in MRI scans before surgery as input for their model. This data was then analyzed to assess how resistant each patient’s brain might be to damage resulting from the tumor removal process.
Previously, predicting cognitive outcomes after treatment was nearly impossible, despite the importance of these outcomes in a patient’s daily life. The information generated by this AI model could assist surgeons in evaluating a patient’s suitability for surgery, potentially preventing vulnerable patients from enduring irreversible neurological disabilities. However, this approach must undergo clinical validation with a large group of patients before it can be widely adopted. Going forward, the researchers plan to integrate more personalized data, such as brain activity, into their predictive model to enhance its accuracy. A more refined model could greatly reduce the risks of neurological impairment following surgery and improve the quality of life for patients recovering from brain tumor treatments.
“This model is based on properties of white matter connections in patient’s brains before surgery,” said Lars Smolders, PhD researcher in the Department of Mathematics and Computer Science, who developed the AI model along with colleagues. “To me, it is fascinating that we can develop a measure of a brain’s vulnerability to damage (inflicted by surgery and/or chemo- and radiotherapy) based only on MRI images.”
Related Links:
TU/e
