Nov 29, 2021

[KBR Covering RNSA 2021] Medical IP showcases AI segmentation technology 바로가기 기자명 Kim Yun-mi Medical IP’s booth at the Technical Exhibits of RSNA 2021 in Chicago, Sunday. CHICAGO, Ill. -- By Kim Yun-mi/Korea Biomedical correspondent - “Medical IP is the only company that owns AI-powered segmentation technology among participants of the annual meeting of Radiological Society of North America (RSNA) 2021. We commercialized a medical metaverse product for the first time in Korea. With the AI segmentation technology completed last year, we expect that we can start patient-tailored precision medicine in earnest.” So said Park Sang-joon, CEO of Medical IP, when he explained the company’s unique technology during an interview with Korea Biomedical Review at the world’s largest radiology forum that opened on Sunday. Park explained that the company used a deep learning algorithm to segment body compositions such as skin, subcutaneous fat, muscle, abdominal visceral fat, internal organs, bones, and central nervous system in CT images. This means that doctors can take out desired body compositions from medical images and quantify them. “Most AI companies work on diagnostic technologies, but we pursue segmentation technology,” Park said. Medical IP started as the first venture company spun off from the Seoul National University Hospital, so the company tried to find which technology would be needed the most at hospitals. As medicine is driven by evidence, the company thought it needed segmentation technology to back up the evidence with numbers, Park said. However, developing segmentation technologies was not accessible, he went on to say. While diagnostic technology distinguishes whether the patient has the disease, segmentation technology should verify every imaging pixel of each body component, which takes a long time, he added. It took five years for Medical IP to complete AI segmentation technology since its foundation in 2015. Medical IP CEO Park Sang-joon speaks during an interview with Korea Biomedical Review on Sunday. Park emphasized that AI segmentation was helpful for patient monitoring, too. Whereas diagnosis ends when a disease is first identified, he noted that monitoring should be continued after treatment. “If you use our technology, you can quantify and monitor whether the disease improved, worsened, or relapsed after treatment.” Medical IP is drawing much attention as it is the only company showcasing “a medical metaverse product” at the Technical Exhibits of RSNA 2021. Park said the product also used the company's original AI segmentation technology. According to Park, a specific segmented body part can be saved as a digital file. This can be put into 3D printing, which is possible only because AI segmentation technology has been completed. Medical IP’s metaverse product was also based on AI segmentation technology, combined with VR and AR, Park said. “AI segmentation technology allows the digitization of all body parts. With this, we can build a ‘digital twin’ of a person on the computer,” Park said. “With this digital twin, we can run various simulations such as surgery that would have been impossible on a real person.” When a doctor maps out an optimal procedure or treatment, they could use this digital twin, which is “the metaverse” that Medical IP pursues, Park said. He repeatedly emphasized that Medical IP was the first Korean company to commercialize a medical metaverse product. As Medical IP’s technology will enable a digital twin or a medical twin, it will advance patient-centered precision medicine, Park said. The company’s AI segmentation is being tried out in actual patient care, he added. “Hospitals have yet to provide evidence by quantifying big data. But after we completed our technology last year, hospitals are eager to buy our solutions,” he said. He added that sales of a new body composition analysis product, which has yet to be authorized, are increasing sevenfold each quarter. Medical IP does not target a specific disease but aims to divide all the body parts through its AI segmentation technology, Park said.