CrossFusion: A Multi-Scale Cross-Attention Convolutional Fusion Model for Cancer Survival Prediction

Rustin Soraki, Huayu Wang, Sitong Liu, Joann G. Elmore, Linda Shapiro
Proceedings of The 9th International Conference on Medical Imaging with Deep Learning, PMLR 315:2906-2921, 2026.

Abstract

Cancer survival prediction from whole slide images (WSIs) relies on capturing prognostic features spanning multiple magnifications, from global tissue architecture to fine-grained cellular morphology. However, current approaches typically face two main limitations: most frameworks focus heavily on single-scale analysis, thereby overlooking the hierarchical context of tissue; meanwhile, existing multi-scale methods often employ simplistic fusion mechanisms (e.g., direct concatenation) that fail to model effective cross-scale interactions. To address these challenges, we propose CrossFusion, a novel multi-scale architecture that introduces a convolutional fusion processor to perform rigorous scale–space integration. Evaluated on six TCGA cancer cohorts, CrossFusion achieves state-of-the-art C-index performance, consistently outperforming both strong single-scale and multi-scale baselines. Furthermore, leveraging domain-specific pathology feature extractors yields additional gains in prognostic accuracy compared to general-purpose backbones.

Cite this Paper

BibTeX
@InProceedings{pmlr-v315-soraki26a, title = {CrossFusion: A Multi-Scale Cross-Attention Convolutional Fusion Model for Cancer Survival Prediction}, author = {Soraki, Rustin and Wang, Huayu and Liu, Sitong and Elmore, Joann G. and Shapiro, Linda}, booktitle = {Proceedings of The 9th International Conference on Medical Imaging with Deep Learning}, pages = {2906--2921}, year = {2026}, editor = {Huo, Yuankai and Gao, Mingchen and Kuo, Chang-Fu and Jin, Yueming and Deng, Ruining}, volume = {315}, series = {Proceedings of Machine Learning Research}, month = {08--10 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v315/main/assets/soraki26a/soraki26a.pdf}, url = {https://proceedings.mlr.press/v315/soraki26a.html}, abstract = {Cancer survival prediction from whole slide images (WSIs) relies on capturing prognostic features spanning multiple magnifications, from global tissue architecture to fine-grained cellular morphology. However, current approaches typically face two main limitations: most frameworks focus heavily on single-scale analysis, thereby overlooking the hierarchical context of tissue; meanwhile, existing multi-scale methods often employ simplistic fusion mechanisms (e.g., direct concatenation) that fail to model effective cross-scale interactions. To address these challenges, we propose CrossFusion, a novel multi-scale architecture that introduces a convolutional fusion processor to perform rigorous scale–space integration. Evaluated on six TCGA cancer cohorts, CrossFusion achieves state-of-the-art C-index performance, consistently outperforming both strong single-scale and multi-scale baselines. Furthermore, leveraging domain-specific pathology feature extractors yields additional gains in prognostic accuracy compared to general-purpose backbones.} }
Endnote
%0 Conference Paper %T CrossFusion: A Multi-Scale Cross-Attention Convolutional Fusion Model for Cancer Survival Prediction %A Rustin Soraki %A Huayu Wang %A Sitong Liu %A Joann G. Elmore %A Linda Shapiro %B Proceedings of The 9th International Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2026 %E Yuankai Huo %E Mingchen Gao %E Chang-Fu Kuo %E Yueming Jin %E Ruining Deng %F pmlr-v315-soraki26a %I PMLR %P 2906--2921 %U https://proceedings.mlr.press/v315/soraki26a.html %V 315 %X Cancer survival prediction from whole slide images (WSIs) relies on capturing prognostic features spanning multiple magnifications, from global tissue architecture to fine-grained cellular morphology. However, current approaches typically face two main limitations: most frameworks focus heavily on single-scale analysis, thereby overlooking the hierarchical context of tissue; meanwhile, existing multi-scale methods often employ simplistic fusion mechanisms (e.g., direct concatenation) that fail to model effective cross-scale interactions. To address these challenges, we propose CrossFusion, a novel multi-scale architecture that introduces a convolutional fusion processor to perform rigorous scale–space integration. Evaluated on six TCGA cancer cohorts, CrossFusion achieves state-of-the-art C-index performance, consistently outperforming both strong single-scale and multi-scale baselines. Furthermore, leveraging domain-specific pathology feature extractors yields additional gains in prognostic accuracy compared to general-purpose backbones.
APA
Soraki, R., Wang, H., Liu, S., Elmore, J.G. & Shapiro, L.. (2026). CrossFusion: A Multi-Scale Cross-Attention Convolutional Fusion Model for Cancer Survival Prediction. Proceedings of The 9th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 315:2906-2921 Available from https://proceedings.mlr.press/v315/soraki26a.html.

Related Material