The potential of AI-assisted gastrectomy with dual highlighting of pancreas and connective tissue

Abstract

Background

Standard gastrectomy with D2 lymph node (LN) dissection for gastric cancer involves peripancreatic lymphadenectomy [1]. This technically demanding procedure requires meticulous dissection within the dissectable layers of connective tissue, while identifying and preserving the pancreas [2]. Our previous study demonstrated the proficiency of Eureka, a surgical artificial intelligence (AI) system, in recognizing both connective tissue and the pancreas [3,4]. Dual highlighting of these structures is expected to reduce surgeon stress by aiding in anatomical identification, thereby ensuring safer and more accurate surgery.

Methods

Connective tissue and the pancreas were highlighted by the surgical AI system in surgical videos on no. 6 (infrapyloric LNs), no. 8 (LNs along the common hepatic artery), and no. 13 (LNs on the posterior surface of the pancreatic head) dissection. These videos were specifically selected as surgeons encountered difficulty in distinguishing the dissectable layers and the pancreatic process.

Results

All videos showed variations of pancreatic morphologies that differed in size and shape. The AI system consistently highlighted the pancreatic process even during initial exploration. Furthermore, it recognized connective tissue, which delineated the appropriate layers for dissection.

Conclusions

The surgical AI system accurately demonstrated dual highlighting of the pancreatic process and connective tissues. Although there are challenges for clinical application, this system can be a valuable tool for anatomical guidance and recognition during surgery, potentially leading to safer and better outcomes.