Physicians and surgeons have long been challenged by the need to understand the volume and grading of meningiomas to develop a therapeutic plan and predict issues related to prognosis. These tumors are the most common form of primary intracranial tumor and are most commonly managed with radiotherapy or surgery.
As technology has advanced and artificial intelligence is being increasingly applied across medical specialties, researchers in neurology and oncology have begun exploring opportunities for helping patients suffering from brain tumors such as meningiomas. A new study, published in Neuroinformatics, has shed light on how artificial intelligence may help in both detecting meningioma lesions and classifying their grades.
The study involved 5,088 patients who had meningiomas as confirmed through histopathological techniques. A deep learning model to automatically find and evaluate meningiomas was applied. The results showed that this automated detection method was both accurate and reliable, detecting meningiomas and classifying their grades in manners consistent with results from manual segmentations.
These new data are promising in terms of the potential to use deep learning to improve patient care, though they do not point to an ability to replace human physicians during the diagnosis and treatment of brain cancer. Instead, the results suggest that technology may be able to support physicians by, for instance, helping with presurgical assessments of tumors and acting as tools to enhance clinical decision making.
Future research will likely focus on specific applications of these types of technologies for patients with meningiomas as well as other types of tumors. More data are required before we can understand the extent to which deep learning algorithms may improve the quality of the care we can provide or the efficiency with which we effectively treat our patients.
Zhang, H. et al. (2020). Deep learning model for the automated detection and histopathological prediction of meningioma. Neuroinformatics, doi: 10.1007/s12021-020-09492-6.