Efficient Whole Slide Pathology VQA via Token Compression

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lyu, Weimin, Hu, Qingqiao, Qi, Kehan, Shi, Zhan, Huang, Wentao, Gupta, Saumya, Chen, Chao
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915530142646272
author Lyu, Weimin
Hu, Qingqiao
Qi, Kehan
Shi, Zhan
Huang, Wentao
Gupta, Saumya
Chen, Chao
author_facet Lyu, Weimin
Hu, Qingqiao
Qi, Kehan
Shi, Zhan
Huang, Wentao
Gupta, Saumya
Chen, Chao
contents Whole-slide images (WSIs) in pathology can reach up to 10,000 x 10,000 pixels, posing significant challenges for multimodal large language model (MLLM) due to long context length and high computational demands. Previous methods typically focus on patch-level analysis or slide-level classification using CLIP-based models with multi-instance learning, but they lack the generative capabilities needed for visual question answering (VQA). More recent MLLM-based approaches address VQA by feeding thousands of patch tokens directly into the language model, which leads to excessive resource consumption. To address these limitations, we propose Token Compression Pathology LLaVA (TCP-LLaVA), the first MLLM architecture to perform WSI VQA via token compression. TCP-LLaVA introduces a set of trainable compression tokens that aggregate visual and textual information through a modality compression module, inspired by the [CLS] token mechanism in BERT. Only the compressed tokens are forwarded to the LLM for answer generation, significantly reducing input length and computational cost. Experiments on ten TCGA tumor subtypes show that TCP-LLaVA outperforms existing MLLM baselines in VQA accuracy while reducing training resource consumption by a substantial margin.
format Preprint
id arxiv_https___arxiv_org_abs_2507_14497
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Efficient Whole Slide Pathology VQA via Token Compression
Lyu, Weimin
Hu, Qingqiao
Qi, Kehan
Shi, Zhan
Huang, Wentao
Gupta, Saumya
Chen, Chao
Computer Vision and Pattern Recognition
Computation and Language
Whole-slide images (WSIs) in pathology can reach up to 10,000 x 10,000 pixels, posing significant challenges for multimodal large language model (MLLM) due to long context length and high computational demands. Previous methods typically focus on patch-level analysis or slide-level classification using CLIP-based models with multi-instance learning, but they lack the generative capabilities needed for visual question answering (VQA). More recent MLLM-based approaches address VQA by feeding thousands of patch tokens directly into the language model, which leads to excessive resource consumption. To address these limitations, we propose Token Compression Pathology LLaVA (TCP-LLaVA), the first MLLM architecture to perform WSI VQA via token compression. TCP-LLaVA introduces a set of trainable compression tokens that aggregate visual and textual information through a modality compression module, inspired by the [CLS] token mechanism in BERT. Only the compressed tokens are forwarded to the LLM for answer generation, significantly reducing input length and computational cost. Experiments on ten TCGA tumor subtypes show that TCP-LLaVA outperforms existing MLLM baselines in VQA accuracy while reducing training resource consumption by a substantial margin.
title Efficient Whole Slide Pathology VQA via Token Compression
topic Computer Vision and Pattern Recognition
Computation and Language
url https://arxiv.org/abs/2507.14497