SlideTIM improves cancer image analysis with spatial patch predictions
Exploiting Spatial Structure for Transductive Few-Shot Classification of Whole-Slide Images
Computer Vision and Pattern Recognition
Summary
Analyzing large cancer tissue images is important but hard and time-consuming. The authors present SlideTIM, a new method that improves patch-level predictions by considering both the image's spatial layout and patch types together. This method makes predictions more accurate, especially when only a few examples are available. When tested on several cancer image datasets, SlideTIM performed better than previous methods.
What this means in practice
- •For medical imaging teams: Improve automated tissue analysis accuracy in cancer diagnosis workflows using limited labeled data.
- •For digital pathology software developers: Enhance pathology software to refine predictions on tissue images by exploiting spatial relationships between image regions.$Commercial implications: Enables building commercial diagnostic tools with better accuracy and fewer annotations by refining patch-level predictions.
Authors
Tiffanie Godelaine, Manon Dausort, Karim El Khoury, Benoît Gérin, Benoît Macq, Christophe De Vleeschouwer
Abstract
Automating the analysis of whole-slide images (WSIs), a key step in cancer diagnosis, has high clinical value, as it can reduce pathologist's workload while improving diagnosis accuracy. Recently, vision-language models have shown promising performance for patch-level classification without requiring any annotation, yet these zero-shot (ZS) predictions remain noisy on fine-grained tasks and must be further refined. A promising direction is to refine all predictions jointly, i.e., a transductive approach. However, most existing methods are not tailored to WSIs. We thus propose SlideTIM, an adaptation to WSIs of the recent transductive approach LC-TIM, which introduces a combined spatial--latent regularizer together with a prior on the patch class distribution. The former enforces spatially and semantically close patches to receive the same predictions, while the prior calibrates the predicted class proportions. Together, they address the complex spatial organization and the strong class imbalance of WSIs. Evaluated on four histology datasets, SlideTIM consistently outperforms all TIM variants, improving the macro-F1 by +8.1pp over the best competing baseline at 1 shot. Compared to the ZS, it raises the macro-F1 by +19.4pp at 1 shot. The code will be made available after submission.