Multimodal Assessment of Pancreatic Cancer Resectability Using Deep Learning

2026-07-15

AI Summary

This study addresses the high inter-observer variability in assessing resectability of pancreatic ductal adenocarcinoma (PDAC), which currently relies heavily on subjective expert interpretation of tumor–vessel relationships in CT imaging. To overcome this limitation, the authors propose a multimodal deep learning framework that integrates 3D contrast-enhanced CT scans with 17 structured clinical variables to automatically classify tumors into the three resectability categories defined by the NCCN guidelines. The method employs Swin-UNETR as the backbone network to simultaneously segment the pancreas, tumor, and critical peripancreatic vessels, and introduces a dynamic multi-task learning strategy that adaptively balances segmentation and classification task weights based on tumor segmentation Dice performance. This anatomy-aware approach achieves significantly improved resectability classification accuracy, reduces inter-expert disagreement, and enhances consistency in clinical decision-making.

Abstract

Accurate determination of pancreatic ductal adenocarcinoma (PDAC) resectability relies on evaluating how the tumor interacts with major peripancreatic vessels on CT imaging, yet expert assessment often shows substantial variability. We introduce a fully automated multimodal deep learning framework that jointly analyzes 3D contrast enhanced CT and structured clinical information to classify patients into the three National Comprehensive Cancer Network (NCCN) resectability categories (upfront resectable, borderline resectable, locally advanced). The approach uses a Swin-UNETR backbone to obtain anatomy aware image representations through auxiliary segmentation of pancreas, tumor, and vascular structures. These features are fused with a compact clinical embedding derived from 17 routinely collected variables and processed by a lightweight classification head. Model training is guided by a dynamic multitask objective that adapts the balance between segmentation and classification based on current tumor Dice performance, promoting feature representations that remain both anatomically informed and discriminative.

Problem

• pancreatic cancer
• resectability
• multimodal assessment
• deep learning
• CT imaging

Innovation

• multimodal deep learning
• Swin-UNETR
• dynamic multitask learning
• pancreatic cancer resectability
• anatomy-aware representation