Multimodal data predicts survival after pancreatic cancer surgery
A Multimodal Dataset for Survival Prediction in Resected Pancreatic Ductal Adenocarcinoma
Computer Vision and Pattern Recognition
Summary
Predicting how long patients survive after surgery for pancreatic cancer is really hard because there aren't many datasets that combine detailed tissue images, genetic information, clinical details, and long-term outcomes. The authors collected data from 302 patients, including tissue slide images, clinical info, and gene mutations, plus survival times. They tested different computer models to predict survival and found that combining clinical and genetic info worked better than using just images. This dataset and their results offer a useful starting point for improving survival predictions in pancreatic cancer.
What this means in practice
- •For hospital data teams: Incorporate multimodal patient data to enhance survival prediction models for pancreatic cancer patients after surgery.
- •For medical imaging developers: Develop improved AI tools that integrate histology images with clinical and genetic data to predict patient outcomes in pancreatic cancer.
Tested on one dataset.
Authors
Anh-Tien Nguyen, Mawuko Tettey, Jacqueline Michelle Metsch, Teresa Zimmer, Niklas Ullrich, Mario Duker, Sandra Rungeling, Kirsten Reuter-Jessen, Tessa Rosenthal, Lena-Christin Conradi, Michael Ghadimi, Alexander Konig, Elisabeth Hessmann, Volker Ellenrieder, Philipp Strobel, Hanibal Bohnenberger, Anne-Christin Hauschild
Abstract
Survival research in pancreatic ductal adenocarcinoma (PDAC) is limited by the scarcity of datasets linking whole-slide histology with clinical, molecular, and long-term outcome data. We present a retrospective single-centre cohort of 302 patients who underwent PDAC resection at University Medical Center Gottingen. The dataset comprises 446 H&E whole-slide images, clinicopathological variables, targeted sequencing data for 154 patients, and overall-survival outcomes. During follow-up, 253 patients died, and the median follow-up was 76 months. To establish initial reference values, we evaluated fourteen survival-prediction configurations using identical five-repetition Monte Carlo cross-validation partitions. Ridge Cox regression using numeric clinicopathological variables achieved a mean concordance of $0.649 \pm 0.042$ and $0.652 \pm 0.046$ after adding KRAS and TP53 mutation status. The image-only attention model achieved $0.603 \pm 0.030$, while multimodal fusion achieved $0.619 \pm 0.025$, the highest concordance among the neural models. These results establish promising initial benchmarks for future research using this pancreas-specific multimodal dataset, paving the way for external validation.