An Innovative Approach to Laparoscopic Organ and Instrument Segmentation using YOLOv8

This study aims to discuss some of the main challenges and difficulties encountered in laparoscopic surgeries. One of the primary difficulties in laparoscopic surgery is the limited field of view and depth perception. The use of a camera and video monitor during the procedure can make it challenging for the surgeon to accurately assess the three-dimensional anatomy and spatial relationships of the surgical site. This can lead to difficulties in identifying and manipulating the targeted structures, increasing the risk of complications. This approach focuses on enhancing real-time decision support for surgical procedures by leveraging artificial intelligence (AI) powered image-guided navigation technology. The focus is on accurately segmenting and localizing the surgical instruments and organs using the YOLOv8 model. Our analysis demonstrated that the YOLOv8 model outperformed the YOLOv7 model, achieving an IoU of 0.86 and mAP of 0.957, thereby affirming its remarkable accuracy and robustness. This presents a promising application for young surgeons, empowering them with real-time identification and spatial localization of intraoperative objects, particularly benefiting minimally invasive surgeries and improving procedural visualization and surgical outcomes by reducing the dependency on immediate object recognition by senior surgeons.