Interactive Whole Slide Images for RL-based Tumour Segmentation
2026-08-17 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors developed a new method using reinforcement learning to find tumors in extremely large medical images called whole-slide images (WSIs). Instead of looking at many small fixed areas, their system treats the entire image like a map to explore by moving, zooming, and marking tumor spots. They tested it on lung cancer images and got similar results to traditional methods but much faster. They also studied how different ways of designing the system affected its performance.
Whole-slide image (WSI)Reinforcement learningTumour segmentationHierarchical multi-resolutionActor-critic architectureProximal policy optimization (PPO)Pulmonary adenocarcinomaPatch-based methodsComputational pathologySequential decision making
Authors
Mohamad Mohamad, Francesco Ponzio, Maxime Gassier, Nicolas Pote, Xavier Descombes
Abstract
Whole-slide image (WSI) analysis remains computationally challenging due to the extremely large spatial resolution of slides and the sparse distribution of tumour regions. We propose an end-to-end reinforcement learning framework for sequential tumour segmentation directly on WSIs. Instead of treating the slide as a predefined collection of candidate patches, we formulate the WSI itself as a hierarchical multi-resolution environment through which an agent navigates using movement, zooming, and tumour selection actions. The agent jointly processes local observations and a global thumbnail representation within an actor-critic architecture trained using proximal policy optimization (PPO). Experiments on pulmonary adenocarcinoma WSIs demonstrate the feasibility of direct sequential tumour segmentation on full slides, achieving comparable coarse segmentation quality relative to patch-based approaches operating at similar magnification levels, while reducing inference time to a few seconds per slide. We further analyse the impact of environment design and action-space granularity. Our results suggest that modelling WSIs as interactive environments provides a promising direction for RL-based computational pathology