Deep Learning-Based Segmentation of Peritoneal Cancer Index Regions from CT Imaging

2026-04-30

AI Summary

This study addresses the current lack of standardized, non-invasive imaging methods for automatically segmenting the 13 abdominal anatomical regions defined by the radiological Peritoneal Cancer Index (rPCI). For the first time, fully automated segmentation of rPCI regions is achieved using nnU-Net and Swin UNETR models evaluated via five-fold cross-validation on 62 expert-annotated and validated CT scans. Results demonstrate that nnU-Net achieves superior performance with a mean Dice coefficient of 0.82—approaching inter-observer agreement (0.88) and significantly outperforming Swin UNETR (0.76). This work advances peritoneal cancer assessment by shifting from invasive laparoscopy toward standardized, non-invasive image analysis, offering clinicians a practical and automated tool for objective disease evaluation.

Abstract

Peritoneal metastases are currently assessed using diagnostic laparoscopy to determine Sugarbaker's Peritoneal Cancer Index (sPCI), which works by dividing the abdomen into 13 regions and scoring each region based on tumor size. A recent consensus study defined 3D regions to facilitate a radiological PCI (rPCI), providing standardized anatomical regions for imaging-based assessment. Despite its clinical value, sPCI is invasive and lacks a standardized imaging counterpart. In this study, we propose a deep learning-based approach to automatically segment the rPCI regions on CT. We evaluate nnU-Net and Swin UNETR on 62 CT scans with rPCI regions manually annotated by three clinical researchers and validated by two expert radiologists. Performance was assessed using five-fold cross-validation with the Dice Similarity Coefficient (Dice), 95th percentile Hausdorff distance and Average Surface Distance. nnU-Net achieved an overall Dice of 0.82, approaching interobserver agreement (0.88) and outperforming Swin UNETR (0.76), with remaining challenges primarily in right flank and small-bowel regions. These results demonstrate feasibility of automated rPCI segmentation, laying the foundation for non-invasive, imaging-based assessment.

Problem

• Peritoneal Cancer Index
• CT imaging
• automated segmentation
• peritoneal metastases
• radiological assessment

Innovation

• deep learning
• rPCI segmentation
• CT imaging
• nnU-Net
• peritoneal metastases