RCSegNeXt: Efficient multi-scale ConvNeXt for rectal cancer segmentation from sagittal MRI scans

Wang Bo, Ting Xue, Leyang Pan, Dingfu Huang, Yi Xiao, Li Fan, Zaiyi Liu, Shiyuan Liu, S Kevin Zhou
Proceedings of The 8th International Conference on Medical Imaging with Deep Learning, PMLR 301:167-182, 2026.

Abstract

Rectal cancer remains a critical global health challenge, significantly contributing to mor-bidity and mortality worldwide. Magnetic resonance imaging (MRI) in a sagittal planeoffers distinct advantages for rectal cancer diagnosis by providing detailed visualization ofthe rectum and its surrounding anatomy. However, automated segmentation of the rectumand associated tumors remains difficult due to tumor heterogeneity and complex anatom-ical structure, which necessitate multi-scale feature extraction. This study proposes RC-SegNeXt, a novel non-uniform pure-convolutional rectal cancer segmentation architecturethat combines shallow anisotropic stages with deep isotropic stages. The anisotropic stagesleverage AniNeXt blocks, designed with customized convolutional kernels and pooling op-erations to address the uneven spatial resolution inherent in MRI data. In the isotropicstages, an IsoNeXt block with a Scale-Aware Integration Module (SAIM) enables efficientmulti-scale feature fusion by directing information flow through constrained pathways. Thisdesign enhances computational efficiency while achieving superior segmentation accuracy.Experiments on two in-house datasets demonstrate the proposed method’s state-of-the-artperformances. Code will be open upon acceptance.

Cite this Paper

BibTeX
@InProceedings{pmlr-v301-bo26a, title = {RCSegNeXt: Efficient multi-scale ConvNeXt for rectal cancer segmentation from sagittal MRI scans}, author = {Bo, Wang and Xue, Ting and Pan, Leyang and Huang, Dingfu and Xiao, Yi and Fan, Li and Liu, Zaiyi and Liu, Shiyuan and Zhou, S Kevin}, booktitle = {Proceedings of The 8th International Conference on Medical Imaging with Deep Learning}, pages = {167--182}, year = {2026}, editor = {Tasdizen, Tolga and Elhabian, Shireen and Summers, Ronald and Chen, Chen and Koch, Lisa and Zhuang, Yan}, volume = {301}, series = {Proceedings of Machine Learning Research}, month = {09--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v301/main/assets/bo26a/bo26a.pdf}, url = {https://proceedings.mlr.press/v301/bo26a.html}, abstract = {Rectal cancer remains a critical global health challenge, significantly contributing to mor-bidity and mortality worldwide. Magnetic resonance imaging (MRI) in a sagittal planeoffers distinct advantages for rectal cancer diagnosis by providing detailed visualization ofthe rectum and its surrounding anatomy. However, automated segmentation of the rectumand associated tumors remains difficult due to tumor heterogeneity and complex anatom-ical structure, which necessitate multi-scale feature extraction. This study proposes RC-SegNeXt, a novel non-uniform pure-convolutional rectal cancer segmentation architecturethat combines shallow anisotropic stages with deep isotropic stages. The anisotropic stagesleverage AniNeXt blocks, designed with customized convolutional kernels and pooling op-erations to address the uneven spatial resolution inherent in MRI data. In the isotropicstages, an IsoNeXt block with a Scale-Aware Integration Module (SAIM) enables efficientmulti-scale feature fusion by directing information flow through constrained pathways. Thisdesign enhances computational efficiency while achieving superior segmentation accuracy.Experiments on two in-house datasets demonstrate the proposed method’s state-of-the-artperformances. Code will be open upon acceptance.} }
Endnote
%0 Conference Paper %T RCSegNeXt: Efficient multi-scale ConvNeXt for rectal cancer segmentation from sagittal MRI scans %A Wang Bo %A Ting Xue %A Leyang Pan %A Dingfu Huang %A Yi Xiao %A Li Fan %A Zaiyi Liu %A Shiyuan Liu %A S Kevin Zhou %B Proceedings of The 8th International Conference on Medical Imaging with Deep Learning %C Proceedings of Machine Learning Research %D 2026 %E Tolga Tasdizen %E Shireen Elhabian %E Ronald Summers %E Chen Chen %E Lisa Koch %E Yan Zhuang %F pmlr-v301-bo26a %I PMLR %P 167--182 %U https://proceedings.mlr.press/v301/bo26a.html %V 301 %X Rectal cancer remains a critical global health challenge, significantly contributing to mor-bidity and mortality worldwide. Magnetic resonance imaging (MRI) in a sagittal planeoffers distinct advantages for rectal cancer diagnosis by providing detailed visualization ofthe rectum and its surrounding anatomy. However, automated segmentation of the rectumand associated tumors remains difficult due to tumor heterogeneity and complex anatom-ical structure, which necessitate multi-scale feature extraction. This study proposes RC-SegNeXt, a novel non-uniform pure-convolutional rectal cancer segmentation architecturethat combines shallow anisotropic stages with deep isotropic stages. The anisotropic stagesleverage AniNeXt blocks, designed with customized convolutional kernels and pooling op-erations to address the uneven spatial resolution inherent in MRI data. In the isotropicstages, an IsoNeXt block with a Scale-Aware Integration Module (SAIM) enables efficientmulti-scale feature fusion by directing information flow through constrained pathways. Thisdesign enhances computational efficiency while achieving superior segmentation accuracy.Experiments on two in-house datasets demonstrate the proposed method’s state-of-the-artperformances. Code will be open upon acceptance.
APA
Bo, W., Xue, T., Pan, L., Huang, D., Xiao, Y., Fan, L., Liu, Z., Liu, S. & Zhou, S.K.. (2026). RCSegNeXt: Efficient multi-scale ConvNeXt for rectal cancer segmentation from sagittal MRI scans. Proceedings of The 8th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 301:167-182 Available from https://proceedings.mlr.press/v301/bo26a.html.

Related Material