Related Experiment Video
Updated: Aug 9, 2026
Retroperitoneal Laparoscopic Debridement and Drainage for Pancreatic Abscess
Published on: March 15, 2024
Automated CT Volumetry of Peripancreatic Collections for Risk Stratification in Acute Pancreatitis: A Multicenter
Siyuan Ouyang1, Yujun Fan1, Fuyao Liu2
1Department of Radiology, The Third Affiliated Hospital of Soochow University, Changzhou, China (S.O., Y.F., Q.J., W.X., J.C., J.Z.); Changzhou Key Laboratory of Medical Imaging, Changzhou, China (S.O., Y.F., Q.J., W.X., J.C., J.Z.); Jiangsu Province Artificical Intelligence for Medical Images Engineering Research Center, Changzhou, China (S.O., Y.F., Q.J., W.X., J.C., J.Z.).
Automated CT volumetry accurately measures peripancreatic collections (PPCs) in acute pancreatitis, enabling faster and more effective risk stratification for organ failure and infection compared to manual methods.
Area of Science:
- Radiology
- Medical Imaging
- Artificial Intelligence in Medicine
Background:
- Extrapancreatic necrosis volume is a key prognostic marker in acute necrotizing pancreatitis.
- Early diagnosis and quantification of peripancreatic collections (PPCs) are challenging and time-consuming with manual segmentation.
- Accurate early risk stratification is crucial for managing acute pancreatitis.
Purpose of the Study:
- To develop and validate an automated CT volumetry tool for PPCs using contrast-enhanced CT (CE-CT).
- To evaluate the utility of automated PPC volume for early risk stratification of organ failure and infection.
- To compare the performance of automated volumetry against manual segmentation and existing scoring systems.
Main Methods:
- Retrospective study including 394 patients with acute pancreatitis.
- Development and internal testing of an automated segmentation model using a primary center cohort (n=303).
- External validation on a cohort of 91 patients from two hospitals.
- Assessment of model performance using Dice coefficients and Pearson correlation.
- Evaluation of automated PPC volume's association with organ failure and infection using logistic regression and ROC analysis.
Main Results:
- Automated segmentation demonstrated excellent agreement with manual assessment in the external test cohort (Dice coefficient: 0.89).
- Automated PPC volume was independently associated with organ failure and infection in both test cohorts (P ≤ 0.006).
- Automated volumetry outperformed the modified CT severity index (mCTSI) in risk stratification for both organ failure and infection.
- The automated tool was significantly faster than manual segmentation (25.2 seconds vs. 12.4 minutes).
Conclusions:
- A deep learning framework enables rapid, fully automated PPC volumetry on CE-CT.
- This tool provides clinically meaningful risk stratification for organ failure and infection in acute pancreatitis.
- Automated PPC volumetry offers a significant advantage over manual methods and mCTSI for early patient assessment.
Related Concept Videos
01:29Chronic Pancreatitis II: Collaborative Care
Assessment:
01:25Acute Pancreatitis I: Introduction