DTC-WSI: Dynamic Token Compression for Whole Slide Images

Tawsifur Rahman, Aliasghar Tarkhan, Rama Chellappa, Alexander Baras

Keywords: Computational pathology, Token compression, Dynamic token reduction, Weakly supervised learning

Abstract: Whole-slide images (WSIs) contain tens of thousands of heterogeneous patches, making transformer-based multiple-instance learning (MIL) computationally expensive due to quadratic attention costs and substantial redundancy in tissue morphology. Existing token-reduction approaches for WSI analysis rely primarily on pruning, which discards information early in training and destabilizes optimization under weak supervision. We propose \textbf{Dynamic Token Compression for Whole-Slide Images (DTC-WSI)}, a token-efficient MIL framework that performs \emph{progressive}, \emph{importance-aware} WSI compression. DTC-WSI integrates a lightweight saliency network with a multi-stage token compressor that combines \emph{bipartite similarity matching} and \emph{soft differentiable pruning} to gradually eliminate redundant or non-diagnostic patches. During training, soft gates enable stable gradient flow, while inference employs deterministic compression for substantial acceleration. This curriculum-style compression preserves discriminative morphology and dramatically reduces computational burden. Across four WSI benchmarks (TCGA-NSCLC, TCGA-BRCA, TCGA-RCC, PANDA), DTC-WSI achieves \textbf{5--10$\times$ token reduction}, \textbf{up to 5.3$\times$ faster inference}, and \textbf{20--40\% lower memory usage}, while improving MIL classification accuracy by \textbf{2--4\%} over state-of-the-art baselines. Our results demonstrate that dynamic token compression is a powerful and scalable alternative to pruning, enabling efficient transformer-based WSI analysis while improving accuracy.

Primary Subject Area: Application: Histopathology

Secondary Subject Area: Detection and Diagnosis

Registration Requirement: Yes

Visa & Travel: Yes

Read CFP & Author Instructions: Yes

Originality Policy: Yes

Single-blind & Not Under Review Elsewhere: Yes

LLM Policy: Yes

Submission Number: 363