Deepnoid Demonstrates Technological Excellence at Global Medical Imaging AI Competition

Deepnoid, a leading first-generation medical AI company in South Korea, announced on October 28 that it secured first place in the Computed Tomography Angiography (CTA) category and second place in the Magnetic Resonance Angiography (MRA) category at the global medical imaging AI competition, the TopBrain Challenge.

The competition was an official registered challenge of the International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI) 2025, held last month in Daejeon. AI models were tasked with identifying and labeling major vessels in brain imaging datasets and evaluating the accuracy of 3D structure segmentation. Unlike previous competitions that focused on the circle of Willis (cerebral arterial circle), this event used a dataset annotated with approximately 40 structures of the brain’s major arteries and veins, allowing AI models to compete based on their performance with this comprehensive data.

The challenge was divided into two categories: CTA and MRA. Teams were ranked based on their performance in these tasks. Evaluation focused not on simple pixel-level accuracy, but rather on ▲vascular connectivity, ▲topological accuracy, and ▲structural completeness. This assessment method goes beyond segmentation accuracy to verify anatomical consistency, aiming to evaluate realistic vascular structures that maintain physiological blood flow in clinical settings.

The 'ARG-DeepNeuro' team, consisting of researchers from Deepnoid’s brain imaging research group, demonstrated the technical excellence of Deepnoid’s proprietary 'ARG-DeepNeuro' model on the global stage through their award-winning performance.

Ko Sungmin, Head of Deepnoid’s Medical R&D Division, stated, "This award is highly significant as it demonstrates that our technology for replicating the intricate anatomical structure of cerebral vessels with AI has gained international recognition. Moving forward, we will continue to develop precise medical AI models that can make tangible contributions to clinical practice by integrating various imaging modalities and clinical data."

Deepnoid is currently working on updates for its digital medical device 'DEEP:NEURO', which shares core technology with the 'ARG-DeepNeuro' model. DEEP:NEURO was designated as an innovative medical technology in 2023 and subsequently obtained a non-reimbursed health insurance code the following year.

This content was produced with the assistance of AI translation services.