TAIPEI (Taiwan News) — A research team from National Taiwan University of Science and Technology has won the 22nd Taiwan National Innovation Award for developing an AI-assisted gynecological surgery system implemented at Tri-Service General Hospital, per Liberty Times.
As advanced medical procedures continue to evolve, improved identification technology is helping reduce the risks associated with high-complexity surgeries. NTUST Professor Guo Jing-ming (郭景明) led the development of a real-time image recognition system that automatically identifies and marks organs and lesions during operations, enhancing the safety of minimally invasive gynecological procedures.
Guo said that while robotic systems such as the da Vinci Surgical System provide high-resolution imaging and stable instrument control, they lack tactile feedback. As a result, surgeons must rely heavily on experience to identify organs, which can present challenges for less experienced physicians or in complex cases.
To address this, Guo’s team compiled a large dataset of surgical images annotated by physicians to train a deep learning model. The resulting system can instantly recognize structures such as the uterus, ovaries, fallopian tubes, intestines, and tumors, highlighting them in color to improve anatomical clarity and lesion boundaries.

According to the research team, the system processes images at 30 frames per second and achieves an organ identification sensitivity of about 90%, approaching the threshold for real-time clinical application.
Guo emphasized that the technology is designed to assist rather than replace surgeons, describing it as “a second pair of eyes” during procedures. The system can also provide real-time alerts in high-risk areas, such as near blood vessels, to improve precision and reduce complications.
The team is continuing to refine the system to ensure stability under varying surgical conditions, including changes in lighting, bleeding, and obstructed views. The technology is also highly scalable and could be extended to other endoscopic and minimally invasive procedures, as well as medical imaging analysis, such as MRI and CT scans.
Looking ahead, the researchers plan to integrate augmented and virtual reality features to support physician training and surgical simulation. The technology has already undergone transfer for commercialization and is moving toward clinical validation to bring AI-assisted tools into operating rooms as a standard support system for surgeons.





