CrossScale-GLIO: Topology-Preserving Vision-Language Alignment of MRI and Whole-Slide Histopathology for Diffuse Glioma
Abstract
Magnetic resonance imaging and histopathology observe the same glioma at radically different scales. We present CrossScale-GLIO, a visual multimodal framework that represents MRI as a tumor-habitat graph and histology as a cell-niche graph, then aligns them with a structure-aware optimal transport objective anchored by diagnostic language. Across paired and external glioma cohorts, CrossScale-GLIO achieved a paired-test subtype macro-F1 of 0.789, IDH AUROC of 0.934, 1p/19q AUROC of 0.884, and MGMT AUROC of 0.802. The subtype gain over feature-only transport was 2.8 percentage points (95% CI: 1.2 to 4.4, adjusted p = 0.0019). Bidirectional patient retrieval reached Recall@1 values of 0.286 and 0.278, and Recall@5 values of 0.621 and 0.608. Pathologists rated 81.2% of high-mass habitat-niche pairs as biologically plausible. Deleting the highest-mass pair reduced correct-class probability by 0.184, compared with 0.049 under random deletion. Degree-preserving graph rewiring reduced subtype macro-F1 by 0.034 and retrieval Recall@1 by 0.090, directly confirming that preserved relational topology drives cross-scale correspondence.
Related
- `Q-niche': topology-preserving niche hypervolumes built using a quantum mechanics modelling analogue.
- What Shapes Brain Alignment of Large Language Models during Dialogue? Insights from Language Production and Comprehension
- Rank-Preserving Alignment Enables Cross-platform Learning and Phenotyping for Single-Cell and Spatial Proteomics
- STAMP as a topology-aware molecular language for fragment-level learning and editing
- A Low-Rank Space Captures Linear Alignment Between LLM and Human Neural Activity During Naturalistic Language Listening
BibTeX
@misc{liu2026crossscaleglio,
title = {{CrossScale-GLIO: Topology-Preserving Vision-Language Alignment of MRI and Whole-Slide Histopathology for Diffuse Glioma}},
author = {Yantong Liu and Zheyu Zhang and Runpeng Liu and Mu Xitang and Seong-Yoon Shin and Hyun-Ae Lee},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.28524},
eprint = {2609.28524},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.28524},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - CrossScale-GLIO: Topology-Preserving Vision-Language Alignment of MRI and Whole-Slide Histopathology for Diffuse Glioma
AU - Yantong Liu
AU - Zheyu Zhang
AU - Runpeng Liu
AU - Mu Xitang
AU - Seong-Yoon Shin
AU - Hyun-Ae Lee
PY - 2026
DA - 2026/09/01
PB - arXiv
DO - 10.48550/arxiv.2609.28524
UR - https://arxiv.org/abs/2609.28524
AB - Magnetic resonance imaging and histopathology observe the same glioma at radically different scales. We present CrossScale-GLIO, a visual multimodal framework that represents MRI as a tumor-habitat graph and histology as a cell-niche graph, then aligns them with a structure-aware optimal transport objective anchored by diagnostic language. Across paired and external glioma cohorts, CrossScale-GLIO achieved a paired-test subtype macro-F1 of 0.789, IDH AUROC of 0.934, 1p/19q AUROC of 0.884, and MGMT AUROC of 0.802. The subtype gain over feature-only transport was 2.8 percentage points (95% CI: 1.2 to 4.4, adjusted p = 0.0019). Bidirectional patient retrieval reached Recall@1 values of 0.286 and 0.278, and Recall@5 values of 0.621 and 0.608. Pathologists rated 81.2% of high-mass habitat-niche pairs as biologically plausible. Deleting the highest-mass pair reduced correct-class probability by 0.184, compared with 0.049 under random deletion. Degree-preserving graph rewiring reduced subtype macro-F1 by 0.034 and retrieval Recall@1 by 0.090, directly confirming that preserved relational topology drives cross-scale correspondence.
ER -