Samsung Medical Center is launching a collaborative effort with universities and corporations to develop real-time ultrasound artificial intelligence (AI) technology for emergency and critical care environments. The project’s primary goal is to create an integrated diagnostic system that medical professionals can use immediately in clinical settings.
On Tuesday, the hospital announced its selection for a new project in the first phase of the government’s multi-agency advanced medical device research and development initiative, focusing on core technology and product development for medical devices. This initiative aims to bolster the domestic medical device industry and strengthen health security capabilities through cutting-edge medical device development.
The hospital plans to invest 5 billion KRW (approximately 3.3 million USD) over five years to jointly develop On-Device-based real-time ultrasound AI technology with Samsung Medison. On-Device technology enables data collection and processing directly on devices such as smartphones, laptops, and appliances, without relying on external cloud servers or internet connections.
Sungkyunkwan University and Ewha Womans University will also participate as research partners. Dr. Park Sung-ji from the Department of Cardiology will lead the hospital’s research team, while Dr. Jeong Myung-jin, a professor in the hospital’s Department of Radiology and director of the Data Science Research Institute, will spearhead the project at Sungkyunkwan University.
This initiative addresses the limitations of current point-of-care ultrasound (POCUS) practices in emergency rooms, intensive care units, and transport settings, where image acquisition and interpretation heavily depend on the examiner’s expertise.
The project’s core focus is developing an integrated diagnostic system that medical staff can use immediately in clinical settings. This system will utilize On-Device Real-Time AI, operating directly within ultrasound equipment without network connectivity.
The diagnostic system will integrate three key components: a scan guide to help obtain standard views, automatic quality assessment (QA) to exclude unsuitable images, and analysis of quantitative and abnormal findings crucial for emergency decision-making.
For cardiac POCUS applications, the system will implement and validate critical functions such as automatic left ventricular ejection fraction (LVEF) evaluation, detection of regional wall motion abnormalities (RWMA), and identification of pericardial effusion and suspected cardiac tamponade.
Dr. Park emphasized that while POCUS is crucial in emergency settings, acquiring standard images and interpreting them accurately remains challenging. It aims to develop an AI-based integrated diagnostic system supporting the entire process from imaging to analysis, backed by clinical evidence to enhance its practical application.
