Deepnoid is moving to fully commercialize its generative medical AI, anchored by regulatory approval of its generative AI-based medical device M4CXR.
M4CXR was trained on more than 10 million chest X-ray images and radiology reports and can detect more than 41 types of abnormal findings. It automatically generates a preliminary findings report in 2.3 seconds.
The device was the first generative AI medical device to receive innovative device designation and has now obtained digital medical device product approval, marking its entry into the commercial stage.
Deepnoid announced the development Monday at a media day for the M4CXR launch held at the Westin Josun Seoul in Sogong-dong.
"Starting with M4CXR, we will expand modalities to chest CT and MRI and advance toward a medical AI agent service," CEO Choi Woo-shik said. "We will reinvent ourselves as a medical AI service company encompassing solutions, infrastructure and agents."
Deepnoid presented M4CXR as a solution to structural problems in the healthcare sector — a shortage of radiologists, reading infrastructure concentrated in the Greater Seoul area and large hospitals, and the heavy burden of interpreting large volumes of chest imaging studies. The company said medical AI has evolved to the point where it can meaningfully assist clinicians with interpretation, delivering practical benefits to frontline hospitals.
"The product approval for M4CXR is a symbolic starting point for the commercialization of generative medical AI — one that creates real-world benefits in the field," Choi said.
Deepnoid said know-how accumulated through earlier products, including DEEP:NEURO and DEEP:CHEST, forms the foundation for deploying M4CXR in clinical settings. Going forward, the company plans to partner with the country's top picture archiving and communication system provider and use its nationwide hospital network to bring M4CXR into the field.
"Commercialization cannot be achieved by model performance alone. Even with superior technology, high costs and power demands can limit adoption," Choi said. "True commercialization requires stable operating infrastructure as well."
He added that the company is conducting a proof-of-concept test applying Furiosa AI's AI semiconductor RNGD (Renegade) to M4CXR.
freiheit@heraldcorp.com
