Detection of Peri-Pancreatic Edema using Deep Learning and Radiomics Techniques

TL;DR

This work targets automated detection of peri-pancreatic edema in pancreatitis by combining pancreas segmentation with two downstream prediction streams: a radiomics-based XGBoost classifier and deep-learning models trained on the pancreas ROI. The authors introduce a public CT dataset of 255 patients with pancreas masks and edema labels, and demonstrate that LinTransUNet achieves solid pancreas segmentation (Dice 80.85%, mIoU 68.73%). Among classifiers, Swin-tiny delivers the best edema-detection performance (recall 98.85%, precision 98.38%), while a radiomics XGBoost model offers a fast, lower-variance alternative (accuracy 79.61%). Overall, the study establishes a practical baseline for peri-pancreatic edema detection and provides valuable data and code for future research in pancreatic imaging.

Abstract

Identifying peri-pancreatic edema is a pivotal indicator for identifying disease progression and prognosis, emphasizing the critical need for accurate detection and assessment in pancreatitis diagnosis and management. This study \textit{introduces a novel CT dataset sourced from 255 patients with pancreatic diseases, featuring annotated pancreas segmentation masks and corresponding diagnostic labels for peri-pancreatic edema condition}. With the novel dataset, we first evaluate the efficacy of the \textit{LinTransUNet} model, a linear Transformer based segmentation algorithm, to segment the pancreas accurately from CT imaging data. Then, we use segmented pancreas regions with two distinctive machine learning classifiers to identify existence of peri-pancreatic edema: deep learning-based models and a radiomics-based eXtreme Gradient Boosting (XGBoost). The LinTransUNet achieved promising results, with a dice coefficient of 80.85\%, and mIoU of 68.73\%. Among the nine benchmarked classification models for peri-pancreatic edema detection, \textit{Swin-Tiny} transformer model demonstrated the highest recall of $98.85 \pm 0.42$ and precision of $98.38\pm 0.17$. Comparatively, the radiomics-based XGBoost model achieved an accuracy of $79.61\pm4.04$ and recall of $91.05\pm3.28$, showcasing its potential as a supplementary diagnostic tool given its rapid processing speed and reduced training time. Our code is available \url{https://github.com/NUBagciLab/Peri-Pancreatic-Edema-Detection}.

Figures (2)

• Figure 1: Proposed workflow for detecting peri-pancreatic edema from CT images using two machine learning approaches: radiomics and deep learning. The process begins with inputting CT images into the LinearTransUnetzhang2022dynamic segmentor for pancreas label generation. Segmented pancreases are in two complementary paths. In the first path, radiomics techniques are applied to the ROI (segmented pancreases) to extract features that are then used by an eXtreme Gradient Boosting (XGBoost) model to make a prediction (Prediction 2) regarding the presence of edema tissue. The second path utilizes the same ROIs with a deep learning-based model (CNN & Transformers) for diagnostic prediction.
• Figure 2: The upper row showcases the original CT scans, while the bottom row displays the extracted pancreas ROI. The images are presented in three different planes: axial, coronal, and sagittal, arranged from left to right.