Abstract
Purpose
Accurate identification of critical anatomical structures, including hernia orifices and Cooper's ligament, is crucial to ensure appropriate mesh placement during transabdominal preperitoneal repair (TAPP). However, the posterior view of the inguinal wall is often less familiar to general surgeons, possibly contributing to the intraoperative misidentification of key anatomical landmarks. We aimed to develop an artificial intelligence (AI)-based model for supporting the recognition of these critical anatomical structures.Methods
We calculated 2,725 images which were extracted from 89 surgical videos of TAPP procedures by three surgeons at a single center. The extracted frames were annotated for the hernia orifices, Cooper's ligament, inferior epigastric vessels, testicular vessels, and vas deferens. Annotation was initially performed by an experienced medical student and subsequently confirmed by two board-certified surgeons. Ten-fold cross-validation was performed to train and assess the AI-based model. Model performance was evaluated using the Intersection over Union (IoU), which quantifies the overlap between AI-generated and ground-truth segmentations on a scale from 0 (no overlap) to 1 (perfect overlap).Results
The mean IoUs for the direct hernia orifices, indirect hernia orifices, Cooper's ligament, inferior epigastric vessels, testicular vessels, and vas deferens were 0.391, 0.325, 0.562, 0.451, 0.555 and 0.554, respectively. The performance was comparable to or exceeding that reported for selected anatomical structures in previous studies.Conclusion
The AI-based model recognized critical anatomical structures during TAPP. Its integration into clinical practice may improve intraoperative recognition of critical anatomical structures and facilitate appropriate mesh placement during TAPP.