Automatic measurement of mesenteric vascular and portal vein parameters via PE-NET in the diagnosis of Crohn's disease

Weize Xu¹Liangfang Zheng¹Kun Zhang^2*Bosheng He^3*

• ¹Affiliated Hospital of Nantong University, Nantong, China
• ²Nantong University, Nantong, China
• ³Nantong First People’s Hospital, Nantong, China

Objective: Vascular changes accompany the course of Crohn's disease (CD). This study evaluated the value of PE-NET for automated measurement of mesenteric vascular and portal vein parameters and explored the performance of PE-NET combined with a support vector machine (SVM) classifier in CD diagnosis. Methods: The automatic vascular segmentation model was trained on computed tomography enterography (CTE) data from our hospital using PE-NET. Segmentation performance was evaluated with sensitivity (SEN), Dice Similarity Coefficient (DSC), and Average Hausdorff Distance (AHD). The model then automatically measured vascular parameters in the classification set, and an SVM classifier was applied based on selected vessel features. Diagnostic performance of PE-NET+SVM was compared with human radiologists and clinical biomarkers (CRP, FCP). The impact of PE-NET+SVM on radiologist reading time was also assessed. Results: The segmentation dataset included CTE data from 54 CD patients and 20 healthy controls, and the classification dataset included 40 CD patients and 45 controls. PE-NET performed well in vascular segmentation of the superior mesenteric artery (SMA), portal vein (PV), and abdominal aorta (AA) in both validation and testing sets. Vascular parameters were automatically extracted by PE-NET, and the mesenteric artery, portal vein, abdominal aorta, and the ratio of portal vein/superior mesenteric artery or abdominal aorta were increased in the testing set, with no statistical difference between automated and manual CTE measurements. An SVM classifier based on these parameters achieved an F1 score comparable to senior radiologists and an AUC of 0.934, exceeding those of clinical biomarkers such as FCP (0.913) and CRP (0.893), demonstrating strong diagnostic potential. The reading time of a junior radiologist was significantly reduced and comparable to a senior radiologist with PE-NET assistance. Conclusion: PE-NET enables automated measurement of mesenteric vascular and portal vein parameters and can assist efficient CD diagnosis.