CrossPan: A Comprehensive Benchmark for Cross-Sequence Pancreas MRI Segmentation and Generalization

Linkai Peng, Cuiling Sun, Zheyuan Zhang, Wanying Dou, Halil Ertugrul Aktas, Andrea M Bejar, Elif Keles, Tamas Gonda, Michael B Wallace, Zongwei Zhou, Gorkem Durak, Rajesh N Keswani, Ulas Bagci
Proceedings of The 9th International Conference on Medical Imaging with Deep Learning, PMLR 315:4182-4216, 2026.

Abstract

Automatic pancreas segmentation is fundamental to abdominal MRI analysis, yet deep learning models trained on one MRI sequence often fail catastrophically when applied to another—a challenge that has received little systematic investigation. We introduce CrossPan, a multi-institutional benchmark comprising 1,386 3D scans across three routinely acquired sequences (T1-weighted, T2-weighted, and Out-of-Phase) from eight centers. Our experiments reveal three key findings. First, cross-sequence domain shifts are far more severe than cross-center variability: models achieving Dice scores above 0.85 in-domain collapse to near-zero ($<$0.02) when transferred across sequences. Second, state-of-the-art domain generalization methods provide negligible benefit under these physics-driven contrast inversions, whereas foundation models like MedSAM2 maintain moderate zero-shot performance through contrast-invariant shape priors. Third, semi-supervised learning offers gains only under stable intensity distributions and becomes unstable on sequences with high intra-organ variability. These results establish cross-sequence generalization—not model architecture or center diversity—as the primary barrier to clinically deployable pancreas MRI segmentation.

Cite this Paper

BibTeX
@InProceedings{pmlr-v315-peng26a, title = {CrossPan: A Comprehensive Benchmark for Cross-Sequence Pancreas MRI Segmentation and Generalization}, author = {Peng, Linkai and Sun, Cuiling and Zhang, Zheyuan and Dou, Wanying and Aktas, Halil Ertugrul and Bejar, Andrea M and Keles, Elif and Gonda, Tamas and Wallace, Michael B and Zhou, Zongwei and Durak, Gorkem and Keswani, Rajesh N and Bagci, Ulas}, booktitle = {Proceedings of The 9th International Conference on Medical Imaging with Deep Learning}, pages = {4182--4216}, 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/peng26a/peng26a.pdf}, url = {https://proceedings.mlr.press/v315/peng26a.html}, abstract = {Automatic pancreas segmentation is fundamental to abdominal MRI analysis, yet deep learning models trained on one MRI sequence often fail catastrophically when applied to another—a challenge that has received little systematic investigation. We introduce CrossPan, a multi-institutional benchmark comprising 1,386 3D scans across three routinely acquired sequences (T1-weighted, T2-weighted, and Out-of-Phase) from eight centers. Our experiments reveal three key findings. First, cross-sequence domain shifts are far more severe than cross-center variability: models achieving Dice scores above 0.85 in-domain collapse to near-zero ($<$0.02) when transferred across sequences. Second, state-of-the-art domain generalization methods provide negligible benefit under these physics-driven contrast inversions, whereas foundation models like MedSAM2 maintain moderate zero-shot performance through contrast-invariant shape priors. Third, semi-supervised learning offers gains only under stable intensity distributions and becomes unstable on sequences with high intra-organ variability. These results establish cross-sequence generalization—not model architecture or center diversity—as the primary barrier to clinically deployable pancreas MRI segmentation.} }
Endnote
%0 Conference Paper %T CrossPan: A Comprehensive Benchmark for Cross-Sequence Pancreas MRI Segmentation and Generalization %A Linkai Peng %A Cuiling Sun %A Zheyuan Zhang %A Wanying Dou %A Halil Ertugrul Aktas %A Andrea M Bejar %A Elif Keles %A Tamas Gonda %A Michael B Wallace %A Zongwei Zhou %A Gorkem Durak %A Rajesh N Keswani %A Ulas Bagci %B Proceedings of The 9th International Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2026 %E Yuankai Huo %E Mingchen Gao %E Chang-Fu Kuo %E Yueming Jin %E Ruining Deng %F pmlr-v315-peng26a %I PMLR %P 4182--4216 %U https://proceedings.mlr.press/v315/peng26a.html %V 315 %X Automatic pancreas segmentation is fundamental to abdominal MRI analysis, yet deep learning models trained on one MRI sequence often fail catastrophically when applied to another—a challenge that has received little systematic investigation. We introduce CrossPan, a multi-institutional benchmark comprising 1,386 3D scans across three routinely acquired sequences (T1-weighted, T2-weighted, and Out-of-Phase) from eight centers. Our experiments reveal three key findings. First, cross-sequence domain shifts are far more severe than cross-center variability: models achieving Dice scores above 0.85 in-domain collapse to near-zero ($<$0.02) when transferred across sequences. Second, state-of-the-art domain generalization methods provide negligible benefit under these physics-driven contrast inversions, whereas foundation models like MedSAM2 maintain moderate zero-shot performance through contrast-invariant shape priors. Third, semi-supervised learning offers gains only under stable intensity distributions and becomes unstable on sequences with high intra-organ variability. These results establish cross-sequence generalization—not model architecture or center diversity—as the primary barrier to clinically deployable pancreas MRI segmentation.
APA
Peng, L., Sun, C., Zhang, Z., Dou, W., Aktas, H.E., Bejar, A.M., Keles, E., Gonda, T., Wallace, M.B., Zhou, Z., Durak, G., Keswani, R.N. & Bagci, U.. (2026). CrossPan: A Comprehensive Benchmark for Cross-Sequence Pancreas MRI Segmentation and Generalization. Proceedings of The 9th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 315:4182-4216 Available from https://proceedings.mlr.press/v315/peng26a.html.

Related Material