AI System Predicts Radiologists’ Eye Movements to Enhance Diagnostic Accuracy and Training
A researcher at the University of Houston has created an AI tool capable of predicting a radiologist’s next point of focus on a chest X-ray—before the radiologist even moves their eyes.
The system, known as MedGaze, aims to enhance radiology education, streamline hospital workflows, and boost the accuracy of AI diagnostics by emulating the visual behavior of expert radiologists.
“We’re not just trying to guess what a radiologist will do next; we’re helping teach machines and future radiologists how to think more like experts by seeing the world as they do,” explained Hien Van Nguyen, associate professor of electrical and computer engineering at UH and lead author of a new study published in Nature Scientific Reports.
Acting as a “digital gaze twin,” MedGaze was developed using thousands of eye-tracking sessions that recorded how experienced radiologists visually analyzed chest X-rays. The system maps not only where radiologists look but also how long they focus and in what order, providing a real-time model of expert visual processing.
Nguyen emphasized that MedGaze sets itself apart from earlier computer vision models by being able to simulate longer and more complex gaze patterns.
“Unlike previous computer vision efforts that focus on predicting scan paths based on specific objects or categories, our approach addresses a broader context of modeling scan path sequences for searching multiple abnormalities in chest x-ray images,” he said. “Specifically, the key technical innovation of MedGaze is its capability to model fixation sequences that are an order of magnitude longer than those handled by the current state-of-the-art methods.”
The University of Houston noted that the tool could help hospitals manage radiology workloads more efficiently, offer deeper insights into how experts navigate complex diagnostic tasks, and strengthen existing AI systems by highlighting image areas that experienced radiologists prioritize.
Though currently focused on chest X-rays, Nguyen and his team plan to expand MedGaze to support other imaging types, such as MRI and CT.
“This opens the door to a unified, AI-driven approach for understanding and replicating clinical expertise across the full spectrum of medical imaging,” Nguyen said.
The research team also included UH graduate students Akash Awasthi and Mai-Anh Vu, along with MD Anderson Cancer Center professors Carol Wu and Rishi Agrawal.