| 英文摘要 |
Surgeons performing Transabdominal pre-peritoneal (TAPP) hernia surgery must locate several critical anatomical structures during dissection of the myopectineal orifice (CVMPO). Creating an Artificial Intelligence (AI) system that can automatically confirm the presence of these structures during the procedure is an essential step for improving procedure safety, standardization, and education with AI technologies. In this study, an initial AI system that automatically recognizes seven critical structures in laparoscopic TAPP video is presented.
In this single-center study, surgical videos from 160 TAPP procedures were utilized to develop and evaluate an AI system that automatically identifies the pubic symphysis, direct hernia orifice, Cooper’s ligament, iliac vein, triangle of Doom, deep inguinal ring, and iliopsoas muscle. To train the system, 130 videos were annotated and verified by two board- certified surgeons. The performance of the system was then assessed in 30 videos of new patients that were not used in the training data.
The system’s performance was assessed in two ways. First, using single-image validation, the AI model was able to detect structures in a single laparoscopic image with an average precision of 51.2%. Second, using video evaluation, the system was able to detect structures throughout the visual inspection phase of the myopectineal orifice, with a mean accuracy of 77.1% and an F-score of 75.4%.
This is the first successful implementation of AI-based automated recognition of crucial anatomical structures in TAPP surgical videos, signifying a major advancement towards AI-assisted TAPP procedures. The system performed strongly in video evaluation. |
本站僅提供期刊文獻檢索。
