MergeSurv: Merging-Based Continual Learning for Survival Analysis on Whole-Slide Images
2026-07-06 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors developed MergeSurv, a method to help computers learn from medical images called Whole Slide Images (WSIs) without forgetting what they've learned before. Instead of retraining the model from scratch for each new type of cancer, they fine-tune a base model on each task and then combine these updates into one unified model. Their approach saves time and protects patient data by not needing to keep old training images. Experiments showed MergeSurv works better than other common ways to keep learning new tasks and avoids forgetting previous knowledge.
Whole Slide ImagesSurvival AnalysisContinual LearningPathologyModel MergingFine-tuningCatastrophic ForgettingTCGA CohortsVision-Language ModelsInference Strategies
Authors
Vu Minh Tran, Doanh C. Bui, Maï K. Nguyen, Khang Nguyen
Abstract
Survival analysis on Whole Slide Images (WSIs) is important in computational pathology for prognosis estimation and treatment planning. However, existing survival models are typically trained independently for each cancer cohort, making continual adaptation computationally expensive for gigapixel-scale WSIs. In this study, we propose MergeSurv, a merging-based continual learning framework for WSI survival analysis. A pathology vision-language foundation model is independently fine-tuned on each task, and the learned parameters are sequentially merged into a unified model without storing previous training data. We further investigate two inference strategies: One-for-All (OFA) and Voting-Expert Aggregation (VEA). Experiments on four TCGA cohorts demonstrate that MergeSurv outperforms naive fine-tuning as well as representative regularization-based and rehearsal-based continual learning methods, while effectively reducing catastrophic forgetting. The results suggest that model merging is a promising direction for scalable and privacy-preserving continual learning in computational pathology.