Using three-dimensional stomach models to identify anatomical landmarks and estimate the resection site in gastric cancer surgery

Background: Estimating the exact resection site and resected gastric volume in various gastrectomies is challenging. This study used three-dimensional (3D) stomach models to simulate different gastrectomies and identify anatomical landmarks for resection sites. Methods: We constructed 3D stomach models from preoperative computed tomography scans of 42 early gastric cancer patients who underwent total gastrectomy, using artificial intelligence–based software. Additional 3D perigastric vascular models were generated for 30 patients to identify landmarks for resection. The lengths and volumes of the resected and remaining stomachs were measured. Results: In both distal gastrectomy (DG) and proximal gastrectomy (PG), the landmark of the resection line on the lesser curvature (LC) side was defined as the lower border of the esophagogastric junction. On the greater curvature (GC) side, the simulated resection line in DG was close to the first branch of the left gastroepiploic artery (18/30), while that in PG corresponded to the last branch of the right gastroepiploic artery (15/30). The mean total volume of the 3D stomach model was 614.5 ± 182.0 mm3. When cutting along the GC and LC at 3, 5, and 7 cm from the pylorus, the remaining mean antrum volumes (%) were 7.8 ± 3.22 mm3, 12.83 ± 4.72 mm3, and 18.72 ± 6.15 mm3, respectively. Conclusions: Preoperative CT-based 3D stomach models were feasible for estimating remnant gastric volume and predicting resection landmarks, which frequently corresponded with conventional resection lines for DG, PG and PPG, despite individual variations. These findings suggest potential value for surgical planning, although further prospective validation is warranted.