mlsanityData to 14 Oct 2026

SLICEChat: Progressive In-Encoder Token Pruning for Whole-Slide Pathology Language Models

arXiv 2609.24894 ↗First posted 21 Sep 2026cs.CV cs.CLcited by 0

On the map: Multimodal LLMs / vision-language models → Medical vision-language models also fits: Visual token pruning and efficiency

also in 2 more
Abstract (arXiv)

Whole-slide pathology images (WSIs) contain gigapixel-scale visual content, creating a major scalability challenge for slide-level multimodal large language models (MLLMs). Existing approaches process thousands of patch tokens and typically apply compression only after slide encoding, leaving multimodal attention computationally expensive. We introduce SLICEChat, a slide-level MLLM that integrates progressive token pruning within a hybrid Mamba--Transformer slide encoder. Mamba layers enable efficient long-range propagation, while Transformer layers preserve global interactions as the sequence is progressively shortened. Between stages, language-supervised, region-aware pruning removes spatially coherent low-utility regions under a controlled keep-rate schedule, producing compact slide representations before multimodal fusion. On SlideBench VQA, SLICEChat achieves 79.84% accuracy on TCGA and 59.09% on BCNB cohorts, outperforming prior slide-level pathology MLLMs, and achieves the highest overall WSI-Bench metrics. It also provides competitive memory usage and the inference latency among the evaluated models. These results demonstrate accurate and computationally efficient multimodal reasoning over gigapixel WSIs.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Medical vision-language models, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 49 of its 60 references that are in the map. First posted on arXiv 21 Sep 2026.

Builds on

49 of 60 references.

Closest ancestors · 8

Same direction, newest first

  1. 2026TC-SSA: Token Compression via Semantic Slot Aggregation for Gigapixel Pathology Reasoning1 Mar
  2. 2026Act Like a Pathologist: Tissue-Aware Whole Slide Image Reasoning28 Feb
  3. 2025PathFLIP: Fine-grained Language-Image Pretraining for Versatile Computational Pathology19 Dec
  4. 2025LoC-Path: Learning to Compress for Pathology Multimodal Large Language Models5 Dec
  5. 2025Efficient Whole Slide Pathology VQA via Token Compression19 Jul
  6. 2024CPath-Omni: A Unified Multimodal Foundation Model for Patch and Whole Slide Image Analysis in Computational Pathology16 Dec
  7. 2024WSI-LLaVA: A Multimodal Large Language Model for Whole Slide Image3 Dec
  8. 2024SlideChat: A Large Vision-Language Assistant for Whole-Slide Pathology Image Understanding15 Oct

Other references · 41

  1. 2017Attention Is All You Needcited by 60,361
  2. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  3. 2023Visual Instruction Tuningcited by 7,869
  4. 2021Masked Autoencoders Are Scalable Vision Learnerscited by 7,403
  5. 2023BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Modelscited by 6,137
  6. 2025Qwen2.5-VL Technical Reportcited by 5,587
  7. 2023Mamba: Linear-Time Sequence Modeling with Selective State Spacescited by 4,718
  8. 2024GPT-4o System Cardcited by 4,588
  9. 2022BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generationcited by 4,322
  10. 2022FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awarenesscited by 3,673
  11. 2021Efficiently Modeling Long Sequences with Structured State Spacescited by 2,465
  12. 2024Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Dualitycited by 1,336
29 more
  1. 2018Attention-based Deep Multiple Instance Learningcited by 1,095
  2. 2024Vision Mamba: Efficient Visual Representation Learning with Bidirectional State Space Modelcited by 1,056
  3. 2022Token Merging: Your ViT But Fastercited by 1,021
  4. 2021DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsificationcited by 864
  5. 2021TransMIL: Transformer based Correlated Multiple Instance Learning for Whole Slide Image Classificationcited by 555
  6. 2024Gated Delta Networks: Improving Mamba2 with Delta Rulecited by 532
  7. 2020Dual-stream Multiple Instance Learning Network for Whole Slide Image Classification with Self-supervised Contrastive Learningcited by 478
  8. 2024LLaVA-PruMerge: Adaptive Token Reduction for Efficient Large Multimodal Modelscited by 402
  9. 2024Prismatic VLMs: Investigating the Design Space of Visually-Conditioned Language Modelscited by 339
  10. 2025Gated Attention for Large Language Models: Non-linearity, Sparsity, and Attention-Sink-Freecited by 283
  11. 2024Rotary Position Embedding for Vision Transformercited by 236
  12. 2024MambaVision: A Hybrid Mamba-Transformer Vision Backbonecited by 224
  13. 2023LongNet: Scaling Transformers to 1,000,000,000 Tokenscited by 204
  14. 2023Quilt-1M: One Million Image-Text Pairs for Histopathologycited by 194
  15. 2025DivPrune: Diversity-based Visual Token Pruning for Large Multimodal Modelscited by 181
  16. 2023Visual Language Pretrained Multiple Instance Zero-Shot Transfer for Histopathology Imagescited by 134
  17. 2023Virchow: A Million-Slide Digital Pathology Foundation Modelcited by 90
  18. 2023Quilt-LLaVA: Visual Instruction Tuning by Extracting Localized Narratives from Open-Source Histopathology Videoscited by 86
  19. 2023Joint Token Pruning and Squeezing Towards More Aggressive Compression of Vision Transformerscited by 81
  20. 2023DiffRate : Differentiable Compression Rate for Efficient Vision Transformerscited by 77
  21. 2025Accelerating Data Processing and Benchmarking of AI Models for Pathologycited by 65
  22. 2025PACT: Pruning and Clustering-Based Token Reduction for Faster Visual Language Modelscited by 49
  23. 2021Predicting Axillary Lymph Node Metastasis in Early Breast Cancer Using Deep Learning on Primary Tumor Biopsy Slidescited by 43
  24. 2023WsiCaption: Multiple Instance Generation of Pathology Reports for Gigapixel Whole-Slide Imagescited by 31
  25. 2024Token Cropr: Faster ViTs for Quite a Few Taskscited by 19
  26. 2025PathAgent: Toward Interpretable Analysis of Whole-slide Pathology Images via Large Language Model-based Agentic Reasoningcited by 19
  27. 2024PA-LLaVA: A Large Language-Vision Assistant for Human Pathology Image Understandingcited by 16
  28. 2024PathInsight: Instruction Tuning of Multimodal Datasets and Models for Intelligence Assisted Diagnosis in Histopathologycited by 6
  29. 2025Hilbert-Guided Sparse Local Attentioncited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026PathSelect: Sequential Token Selection for Whole Slide Pathology
  2. 2026Simple Token-Efficient Vision-Language Model for Case-level Pathology Synoptic Report Generation
  3. 2026MLLM-HWSI: A Multimodal Large Language Model for Hierarchical Whole Slide Image Understanding
  4. 2025MPath: Multimodal Pathology Report Generation from Whole Slide Images
  5. 2024PathAlign: A vision-language model for whole slide images in histopathology
  6. 2026Benchmarking Vision-Language Models for Automated Pathology Diagnosis and Report Generation
  7. 2025LanGuSTE: Language-Guided Coarse-to-Fine Patch Selection for Efficient Whole Slide Image Analysis
  8. 2025Democratising Pathology Co-Pilots: An Open Pipeline and Dataset for Whole-Slide Vision-Language Modelling