Research map / Architectures and efficiency / Attention and transformers
From Multi-Resolution Cells to Gigapixel Whole Slide Images Foundation Model for Computational Pathology
Basit Alawode, Moshira Abdalla, Dwarikanath Mahapatra, Muhammad Muzammal Naseer, Sajid Javed
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add multi-resolution inputs by concatenation (ablation) | HIPT-style hierarchy · PANDA | balanced accuracy: 0.704 -> 0.762 | Improved |
| add CCRA module (ablation) | MRPT · PANDA | balanced accuracy: 0.762 -> 0.823 | Improved |
| add ACT and [CLS] concat token fusion (ablation) | MRPT · PANDA | balanced accuracy: 0.823 -> 0.848 | Improved |
| add 3-stage hierarchical SSL (ablation) | MRPT · PANDA | balanced accuracy: 0.848 -> 0.866 | Improved |
| use consecutive CCRA instead of distant pairing 10x-40x (ablation) | MRPT · PANDA | balanced accuracy: 0.824 -> 0.866 | Improved |
| use all three resolutions instead of 10x+20x (ablation) | MRPT · PANDA | balanced accuracy: 0.852 -> 0.866 | Improved |
| use CCRA instead of averaged [CLS] token fusion (ablation) | MRPT · PANDA | balanced accuracy: 0.848 -> 0.862 | Improved |
| use CCRA instead of all-token attention (ablation) | MRPT · PANDA | balanced accuracy: 0.850 -> 0.862 | Improved |
| increase ViT10x embedding size to 768 (ablation) | MRPT · PANDA | balanced accuracy: 0.866 -> 0.846 | No clear effect |
| increase CCRA layers K to 2 (ablation) | MRPT · PANDA | balanced accuracy: 0.866 -> 0.843 | No clear effect |
| train and test on synthetic pyramid from 20x only (ablation) | MRPT · PANDA | balanced accuracy: 0.866 -> 0.723 | Hurt |
| replace HIPT with MRPT under matched data | MRPT · Camelyon17 | F1: 0.782 -> 0.883 | Improved |
Closest papers by meaning
- Multi-Resolution Pathology-Language Pre-training Model with Text-Guided Visual Representation (CVPR 2025 · similarity 0.91)
- MLLM-HWSI: A Multimodal Large Language Model for Hierarchical Whole Slide Image Understanding (CVPR 2026 · similarity 0.90)
- PathVQ: Reforming Computational Pathology Foundation Model for Whole Slide Image Analysis via Vector Quantization (NeurIPS 2025 · similarity 0.90)
- Rethinking Transformer for Long Contextual Histopathology Whole Slide Image Analysis (NeurIPS 2024 · similarity 0.89)
- SlideChat: A Large Vision-Language Assistant for Whole-Slide Pathology Image Understanding (CVPR 2025 · similarity 0.88)
- Generalizable Whole Slide Image Classification with Fine-Grained Visual-Semantic Interaction (CVPR 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
