First place in the CTA category and second place in the MRA category demonstrate the technical excellence of ARG-DeepNeuro, a cerebrovascular segmentation model
“An achievement that internationally recognizes our technology for precisely segmenting the complex anatomy of cerebral vessels with AI”
DEEPNOID, one of Korea’s first-generation medical AI companies (CEO Woo Sik Choi), announced on October 28, 2025, that it took first place in the CTA category and second place in the MRA category at the TopBrain Challenge, a global medical imaging AI competition.
The TopBrain Challenge (TopBrain Segmentation Challenge for Whole Brain Vessel Anatomy) is an official challenge of MICCAI 2025 (International Conference on Medical Image Computing and Computer Assisted Intervention), held in Daejeon, Korea, in September 2025. It evaluates how accurately AI models identify and label major vessels in brain imaging datasets and segment their 3D structures. Unlike previous competitions that focused on datasets covering the circle of Willis, the TopBrain Challenge uses a dataset in which more than 40 structures of the major arteries and veins across the whole brain are annotated.
The challenge consists of two categories, computed tomography angiography (CTA) and magnetic resonance angiography (MRA), and teams are ranked based on their performance in each. Rather than simple pixel-level accuracy, the evaluation focuses on vessel connectivity, topological accuracy, and structural completeness. This approach assesses anatomical consistency, not just segmentation accuracy, with the aim of evaluating realistic vascular structures in which clinically meaningful blood flow is preserved.
DEEPNOID’s ARG-DeepNeuro team, made up of Dahye Lee and Kwanseok Oh, researchers in DEEPNOID’s Brain Imaging Research, demonstrated the technical excellence of ARG-DeepNeuro, the company’s in-house model, on the global stage.
“This award is meaningful because it shows international recognition of our technology for reproducing the detailed anatomy of cerebral vessels with AI,” said Sungmin Ko, Head of Medical R&D Division at DEEPNOID. “We will continue to develop precise medical AI models that make a real contribution to clinical practice by linking a variety of imaging modalities with clinical data.”
DEEPNOID is updating DEEP:NEURO, a digital medical device that shares its core technology with the ARG-DeepNeuro model. DEEP:NEURO was designated as an Innovative Medical Technology in 2023 and obtained a non-reimbursable billing code under Korea’s National Health Insurance in 2024.
###
