Research map / Vision-language and multimodal / Cross-modal learning
PS3: A Multimodal Transformer Integrating Pathology Reports with Histology Images and Biological Pathways for Cancer Survival Prediction
Manahil Raza, Ayesha Azam, Talha Qaiser, Nasir Rajpoot
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 |
|---|---|---|---|
| use average report length instead of 90th percentile for text prototype count (ablation) | PS3 · TCGA six cohorts (average) | C-Index: 0.691 -> 0.699 | Improved |
| use PLIP instead of QUILT-Net features (ablation) | PS3 · TCGA six cohorts (average) | C-Index: 0.676 -> 0.699 | Improved |
| add pathology report modality to histology and pathways (ablation) | PS3 · TCGA six cohorts (average) | C-Index: 0.653 -> 0.699 | Improved |
| add pathway modality to histology and text (ablation) | PS3 · TCGA six cohorts (average) | C-Index: 0.644 -> 0.699 | Improved |
| add histology modality to pathways and text (ablation) | PS3 · TCGA six cohorts (average) | C-Index: 0.687 -> 0.699 | Improved |
| use full attention fusion instead of late fusion (ablation) | PS3 · TCGA six cohorts (average) | C-Index: 0.669 -> 0.699 | Improved |
| use joint fusion instead of hierarchical fusion (ablation) | PS3 · TCGA six cohorts (average) | C-Index: 0.660 -> 0.699 | Improved |
| add learnable random embeddings e_r (ablation) | PS3 · TCGA six cohorts (average) | C-Index: 0.688 -> 0.699 | Improved |
| use multiple MLPs instead of one shared MLP (ablation) | PS3 · TCGA six cohorts (average) | C-Index: 0.690 -> 0.699 | Improved |
| use pathway prototypes instead of plain gene expression MLP | unimodal gene model · TCGA six cohorts (average) | C-Index: 0.630 -> 0.652 | Improved |
| use diagnostic prototypes instead of ABMIL on text | unimodal text model · TCGA six cohorts (average) | C-Index: 0.592 -> 0.615 | Improved |
| use histological prototypes instead of TransMIL | unimodal histology model · TCGA six cohorts (average) | C-Index: 0.623 -> 0.588 | Hurt |
Closest papers by meaning
- Multimodal Prototyping for cancer survival prediction (ICML 2024 · similarity 0.92)
- Modeling Dense Multimodal Interactions Between Biological Pathways and Histology for Survival Prediction (CVPR 2024 · similarity 0.92)
- Prototypical Information Bottlenecking and Disentangling for Multimodal Cancer Survival Prediction (ICLR 2024 · similarity 0.89)
- Robust Multimodal Survival Prediction with Conditional Latent Differentiation Variational AutoEncoder (CVPR 2025 · similarity 0.87)
- Interpretable Vision-Language Survival Analysis with Ordinal Inductive Bias for Computational Pathology (ICLR 2025 · similarity 0.87)
- TumorChain: Interleaved Multimodal Chain-of-Thought Reasoning for Traceable Clinical Tumor Analysis (ICLR 2026 · similarity 0.86)
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.
