Automated Prediction of Postoperative Pancreatic Fistula Using Preoperative Computed Tomography
Researchers have developed a deep learning pipeline to predict postoperative pancreatic fistula (POPF) using preoperative CT scans. The system automates the process from pancreatic segmentation to classification, aiming to estimate and stratify POPF risk. Evaluations using various 3D convolutional neural network architectures showed promising predictive performance, offering a valuable tool for preoperative decision-making in pancreatic surgery. AI
IMPACT This research offers a potential new tool for surgeons to better assess patient risk before complex operations.