Photo 1. DEEPNOID’s M4CT team (from left: Kyoyun Choi, Jonggwon Park, and Byungmu Yoon)
- First place in two categories and third place in one category at an official MICCAI challenge
- Demonstrates the technical excellence of M4CT, a 3D vision-language model that shares its technological roots with M4CXR
- Proves the strength of multimodal generative AI technology in interpreting 3D CT images as well as X-ray images
DEEPNOID, one of Korea’s first-generation medical AI companies (CEO Woo Sik Choi), announced on September 29, 2025, that it won first place in two categories and third place in one category at the VLM3D Challenge, a global medical imaging AI competition.
The VLM3D Challenge is an official challenge of MICCAI 2025 (International Conference on Medical Image Computing and Computer Assisted Intervention), held in Daejeon, Korea, on September 24, 2025. Participants compete on the innovation of their 3D vision-language model technologies using CT-RATE, the world’s largest open dataset of paired chest CT scans and radiology reports. The challenge consists of four categories, and teams are ranked based on their performance on the tasks in each category.
DEEPNOID’s M4CT team, made up of DEEPNOID researchers Jonggwon Park, Kyoyun Choi, and Byungmu Yoon, competed in three categories with M4CT, the company’s in-house solution.
The team won first place in Self-Supervised Multi-Abnormality Localization and Multi-Abnormality Classification, and third place in Radiology Report Generation. With these results, DEEPNOID demonstrated the technical excellence of M4CT, its in-house 3D vision-language model, on the global stage.
As of September 2025, DEEPNOID was conducting a multicenter clinical trial of M4CXR, a digital medical device that shares its technological foundation with M4CT. M4CXR interprets 41 types of lesions on chest X-ray images within seconds and automatically generates AI-generated draft reports. At the time, DEEPNOID was aiming to obtain regulatory approval for M4CXR as early as the end of 2025.
“This achievement proves that DEEPNOID’s multimodal generative AI technology excels not only in 2D X-ray images but also in 3D images such as CT,” said Jonggwon Park, leader of DEEPNOID’s M4CT team. “We will also focus on research and development toward commercialization so that M4CT can be used effectively in clinical practice.”
###
