Laparoscopic Colorectal Surgery with Anatomical Recognition with Artificial Intelligence Assistance for Nerves and Dissection Layers

Abstract Background In digestive tract surgery, dissection of an avascular space consisting of loose connective tissue (LCT) appearing by countertraction improves oncological outcomes and reduces complications. 1–3 Kumazu et al. 4 described a deep learning approach that automatically segments LCT to help surgeons. 4 During left colorectal surgery, lumbar splanchnic, hypogastric, and pelvic visceral nerve injuries cause sexual dysfunction and/or urinary issues. 5 As nerve preservation is critical for functional preservation, the AI model Kumazu reported is named Eureka (Anaut Inc., Tokyo, Japan) and was developed to separate nerves automatically. The educative efficacy of intraoperative nerve visualization has been described. 6 Artificial intelligence (AI) assisted navigation is expected to aid in the anatomical recognition of nerves and the safe dissection layers surrounding nerves in the future. Methods We used Eureka as an educational tool for surgeons’ training during laparoscopic colorectal surgery. The laparoscopic system used was Olympus VISERA ELITE3 (Tokyo, Japan). Results Total mesorectal excision (TME) was safely performed with nerve preservation. No postoperative complications occurred. Automatic segmentation and highlighting of LCT in the dissected layers, lumbar splanchnic, hypogastric, and pelvic visceral nerves (S3, S4), were performed in real time. Conclusions In colorectal cancer surgery, the nerves are vital anatomical structures serving as landmarks for dissection. Lumbar splanchnic, hypogastric, and pelvic visceral nerve injuries (S3, S4) cause sexual dysfunction or urinary disorders. 5 Nerve preservation is important for functional preservation. AI-assisted navigation methods are noninvasive, user-friendly, and expected to improve in accuracy in the future. They have the potential to develop nerve-guided TME.