AI-powered brain scanning model recognizes stroke, brain tumors and aneurysms

Such delays often delay the start of treatment and can worsen the prognosis for patients, since MRI plays a key role in detecting and monitoring many brain diseases, including strokes, tumors and aneurysms. AI can reduce the burden on radiology departments by automatically sorting images and speeding up the preparation of reports.

To test the effectiveness, the developed model was first trained to determine whether a scan was normal or contained pathology. The results were comparable to those of experienced radiologists. The system was then tested to detect specific diseases – stroke, multiple sclerosis and brain tumors – using new MRI images that were not part of the training set. The model successfully coped with this task.

Typically, AI is trained on large data sets that are manually labeled by experts, which requires significant resources. To avoid these difficulties, the researchers created a model that could learn on its own using more than 60,000 available brain MRIs along with relevant medical reports. The scientists also configured the model so that it can find and show similar cases based on an uploaded MRI image or a text query—for example, “glioma” (a type of brain tumor). This can be useful both for clarifying the diagnosis and for training specialists.

The results of the study demonstrate that the model can be used directly in the MRI process: it is capable of detecting anomalies in real time and helping doctors make decisions. AI can suggest possible findings, point out errors in conclusions, or select similar cases from archives. All this speeds up the diagnostic process, reduces waiting time for results and ultimately improves the quality of patient treatment.