Pretrained AI could help clinics analyze brain MRIs with few labeled scans
When it comes to medical imaging, a chatbot doesn't know its backside from its elbow. Or, at least, it can't tell a spine from a knee. But if a Boston University research team is successful, a smarter AI system for evaluating brain scan images might someday help clinicians spot diseases sooner, saving lives while cutting costs.
A team of researchers at Boston University has developed an artificial intelligence system that could revolutionize how brain MRIs are analyzed in clinics. The AI model is trained to recognize general patterns in MRI scans, allowing it to be repurposed for various tasks later on. This means that neuroradiologists could use the AI tool to identify lesions or early signs of dementia, even with limited labeled data.
Traditionally, medical imaging required extensive labeled data to train models from scratch, which was both time-consuming and costly. However, the BU team's approach uses a technique called masked autoencoder, which allows the AI to learn from the available data and still perform well despite limited labeled examples.
The AI model is based on a transformer architecture, similar to the one used in ChatGPT. While transformers are excellent at processing local features and global representations, they have not been widely used in medical imaging before. The BU team's work suggests that transformers could be a game-changer for medical imaging analysis, potentially providing faster and more accurate diagnoses for patients.
The implications of this technology could be life-saving, as early detection of conditions like strokes could lead to immediate treatment and improved patient outcomes. The team's research, which has been published in Frontiers in Artificial Intelligence, demonstrates the potential of this AI system to improve diagnostic accuracy and timeliness in clinical settings.
Written by urgent.news from Medical Xpress's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
