TLDR: South Korean AI specialist Deepnoid has received regulatory approval from the Ministry of Food and Drug Safety (MFDS) to commence clinical testing of its generative AI radiology device, M4CXR. Designed to automatically generate medical imaging-based disease diagnoses and reading reports from chest X-rays, M4CXR aims to significantly enhance the efficiency and speed of medical diagnostics, addressing the shortage of radiologists.
Deepnoid, a leading South Korean artificial intelligence company, has achieved a significant milestone with the regulatory approval from the Ministry of Food and Drug Safety (MFDS) to initiate clinical trials for its innovative generative AI radiology device, M4CXR. This approval, granted on August 19, 2025, paves the way for what Deepnoid describes as the first large-scale clinical test of a radiology tool built on a large language model (LLM) rather than a traditional neural network.
M4CXR is engineered to revolutionize medical imaging by automatically generating disease diagnoses and comprehensive reading reports from chest X-rays within seconds. The device has been trained on an extensive dataset of over 10 million X-rays and their corresponding reports, enabling it to identify and flag 41 distinct types of lesions. This capability is expected to dramatically improve diagnostic speed, particularly in emergency situations where timely interpretation is critical.
Kim Sung-hyun, chief director of the Human Medical Imaging and Intervention Center, highlighted the pressing need for such technology, stating, ‘Chest X-rays are the most frequently performed imaging exam, but the shortage of radiologists has limited timely interpretation.’ He added that M4CXR ‘can deliver preliminary results instantly, improving diagnostic speed in emergencies and helping radiologists achieve greater accuracy.’
Unlike existing AI medical devices that primarily identify issues, M4CXR goes a step further by offering solutions and diagnoses to those problems. Choi Hyun-seok, a radiologist and Chief Medical Officer at Deepnoid, likened M4CXR’s efficiency to ‘having a first-year trainee doctor reading alongside you,’ emphasizing its potential to assist non-specialists in achieving diagnostic accuracy closer to that of radiologists.
The clinical study for M4CXR will be conducted in collaboration with prominent medical institutions, Kangbuk Samsung Hospital and Boramae Medical Center. The primary objective of these trials is to rigorously prove the device’s accuracy and safety, a crucial step towards its broader adoption in clinical settings.
Deepnoid’s ambitions extend beyond domestic approval. The company plans to launch the trial next month and then pursue product approval and market entry through Korea’s fast-track pathway for non-reimbursed use, before seeking insurance coverage. Furthermore, Deepnoid is actively pursuing global expansion, having established a subsidiary in Dubai to explore Middle Eastern markets and eyeing eventual entry into the U.S. market. The company is also strategically focusing on Japan as a key testing ground for its AI technology, aiming to validate its solutions before tackling the U.S. FDA 510(k) clearance.
Earlier studies have already demonstrated M4CXR’s robust performance. A study titled ‘Comparison of Chest X-ray Interpretation Performance between Deepnoid’s M4CXR and OpenAI’s ChatGPT,’ published in the Journal of Clinical Medicine, revealed that M4CXR consistently outperformed ChatGPT across five metrics, including accuracy and location accuracy. This underscores M4CXR’s suitability for specialized medical environments compared to general-purpose AI models.
Also Read:
- South Korea’s Privacy Watchdog Releases Comprehensive Generative AI Guidelines
- Generative AI Adoption Soars Among South Korean Workforce, Central Bank Survey Reveals Modest Productivity Gains
Deepnoid, established in 2008, is an AI specialist company with a portfolio of AI solutions across healthcare, security, and IT manufacturing. Its existing medical AI products, such as ‘DEEP:NEURO’ for cerebrovascular diseases and ‘DEEP:CHEST’ for real-time chest disease detection, have already received MFDS approval and are in use in local hospitals. The development of M4CXR aligns with the MFDS’s proactive stance, which released the ‘world’s first’ review guidelines for generative AI medical devices in January 2025, signaling a supportive regulatory environment for such innovations.
