European journal of surgical oncology : the journal of the European Society of Surgical Oncology and the British Association of Surgical Oncology · 2025
AI-assisted precise identification and segmentation of parathyroid glands in multi-approach endoscopic and robotic thyroid surgeries.
СтатьяVideo-Audio MediaНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Liu Jin-Yuan, Zhang Si-Cheng, Ma Qian, Zhang Chao-Ming, Zhang Long-Yue, Zhou Yu-Qiu, Cai Yong-Cong, Ma Lin-Jie, Hu Bin-Tao, Yang Yan, Wang Xu, Shui Chun-Yan, Sun Rong-Hao, Li Chao
Аннотация
BACKGROUND: With the growing demand for cosmetically appealing, scar-free thyroidectomy procedures driving rapid advances in endoscopic and robotic technology, protecting the parathyroid glands (PGs) during remote-access thyroid surgery is essential to minimize complications and ensure favorable outcomes. However, current methods rely primarily on the surgeon's experience and are limited in complex environments. In this study, we developed a comprehensive artificial intelligence (AI) model for precise PG identification and segmentation during multi-approach remote-access thyroid surgery.
METHODS: A total of 210 surgical videos from Sichuan Cancer Hospital, encompassing multiple remote-access approaches, were used to develop AI models. We trained and evaluated five AI models and selected the best-performing YOLOv11 model for further development of our intelligent system, SmartThyroid. The differences in performance between SmartThyroid and surgeons in terms of PG recognition rates, time, and duration were analyzed.
RESULTS: SmartThyroid demonstrated excellent performance, achieving a mean Dice of 0.873, a mean IoU of 0.804, and an AP50 of 0.974. In the test group, SmartThyroid outperformed all surgeons in initial recognition time. It showed higher PG recognition rates and longer duration compared to junior surgeons, while performing similarly to senior surgeons. Furthermore, SmartThyroid significantly reduced initial recognition time and improved duration for all surgeons. Junior surgeons also showed enhanced PG recognition rates.
CONCLUSIONS: SmartThyroid is an innovative AI model for precise PG identification and segmentation during multi-approach remote-access thyroid surgeries. It enhances the accuracy and efficiency of PGs identification in testing.
METHODS: A total of 210 surgical videos from Sichuan Cancer Hospital, encompassing multiple remote-access approaches, were used to develop AI models. We trained and evaluated five AI models and selected the best-performing YOLOv11 model for further development of our intelligent system, SmartThyroid. The differences in performance between SmartThyroid and surgeons in terms of PG recognition rates, time, and duration were analyzed.
RESULTS: SmartThyroid demonstrated excellent performance, achieving a mean Dice of 0.873, a mean IoU of 0.804, and an AP50 of 0.974. In the test group, SmartThyroid outperformed all surgeons in initial recognition time. It showed higher PG recognition rates and longer duration compared to junior surgeons, while performing similarly to senior surgeons. Furthermore, SmartThyroid significantly reduced initial recognition time and improved duration for all surgeons. Junior surgeons also showed enhanced PG recognition rates.
CONCLUSIONS: SmartThyroid is an innovative AI model for precise PG identification and segmentation during multi-approach remote-access thyroid surgeries. It enhances the accuracy and efficiency of PGs identification in testing.