Anatomic Variation Of The Inferior Mesenteric Artery Detected Using 3-dimensional Residual Network

Objective: 1.To investigate the variation of the inferior mesenteric artery by abdominal enhanced CT.2.To explore the practical value of 3-dimensional residual network in detecting common variation of the inferior mesenteric artery.3.To study the correlation between common variation of the inferior mesenteric artery and patients’ clinical characteristic information.4.To analyze the significance of anatomical variation of the inferior mesenteric artery in abdominal surgery.Methods: 1.A total of 220 patients who underwent “laparoscopic left hemicolectomy and laparoscopic radical resection of rectal cancer”in the Department of Gastrointestinal Surgery,The First Affiliated Hospital of University Of South China from August 2017 to December2019 were retrospectively selected.2.The variation of the inferior mesenteric artery were classified and labeled by reading preoperative abdominal enhanced CT reconstruction images of patients.3.An artificial intelligence model with three modules of image preprocessing-feature extraction-vascular variation classification was built for data learning and detection,and the detection performance of the model was evaluated.4.Correlation analysis was conducted between common vascular variation types of patients and clinical characteristics such as gender,age,height,weight,body mass index,tumor stage,distance D(three-dimensional straight-line distance from the inferior mesenteric artery root to abdominal aortic bifurcation),distance d(three-dimensional straight-line distance from the inferior mesenteric artery root to the origin point of the left colic artery).Results:According to Murono classification,there were 113 cases of type I(the left colic artery arose independently from the sigmoid artery),accounting for 51.36%;Type II(the left colic artery and the sigmoid artery arose from the inferior mesenteric artery at the same point):70 patients,31.82%;Type III(the left colic artery and the sigmoid artery had a common trunk): 34 patients,15.45%;Type IV(there was a deficit of the left colic artery):3 patients,1.36%.There was no statistical significance between the common vascular variation types of patients and the clinical characteristics such as gender,age,height,weight,body mass index,tumor stage and distance D(three-dimensional straight-line distance from the inferior mesenteric artery root to abdominal aortic bifurcation),but there was statistical significance between the common vascular variation types and distance d(three-dimensional straight-line distance from the inferior mesenteric artery root to the origin point of the left colic artery)(P < 0.05).In the artificial intelligence model,the average F-measure,precision and recall of the five groups of type I,type II,type III were 0.82,90% and 75%;0.72,77% and 68%;0.62,46% and94% respectively.The artificial intelligence model average F-measure,precision and recall of the five groups are 0.75,71%,79%.Conclusion: 1.The results of this study were similar to those of the traditional method to study the variation of the inferior mesenteric artery,in which type I variation was the most.2.This 3-dimensional residual network has a good detection efficiency for common variation of the inferior mesenteric artery.3.The common vascular variation types were unrelated to the patient’s clinical characteristics such as gender,age,height,weight,body mass index,tumor stage,distance D(three-dimensional straight-line distance from the inferior mesenteric artery root to abdominal aortic bifurcation),but correlated with distance d(three-dimensional straight-line distance from the inferior mesenteric artery root to the origin point of the left colic artery).