# Style Curriculum Learning for Robust Medical Image Segmentation

arXiv: https://arxiv.org/abs/2108.00402 · HTML: https://mlsanity.com/paper/2108.00402/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Builds on 17 of its 21 references that are in the map.
- First posted on arXiv 1 Aug 2021.

## Builds on (most cited first, 17 of 17)

- [U-Net: Convolutional Networks for Biomedical Image Segmentation](https://mlsanity.com/paper/1505.04597/) (2015)
- [Explaining and Harnessing Adversarial Examples](https://mlsanity.com/paper/1412.6572/) (2014)
- [mixup: Beyond Empirical Risk Minimization](https://mlsanity.com/paper/1710.09412/) (2017)
- [CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features](https://mlsanity.com/paper/1905.04899/) (2019)
- [Multimodal Unsupervised Image-to-Image Translation](https://mlsanity.com/paper/1804.04732/) (2018)
- [Generalizing to Unseen Domains via Adversarial Data Augmentation](https://mlsanity.com/paper/1805.12018/) (2018)
- [On The Power of Curriculum Learning in Training Deep Networks](https://mlsanity.com/paper/1904.03626/) (2019)
- [Demystifying Neural Style Transfer](https://mlsanity.com/paper/1701.01036/) (2017)
- [Translating and Segmenting Multimodal Medical Volumes with Cycle- and Shape-Consistency Generative Adversarial Network](https://mlsanity.com/paper/1802.09655/) (2018)
- [Curriculum Adversarial Training](https://mlsanity.com/paper/1805.04807/) (2018)
- [Open Compound Domain Adaptation](https://mlsanity.com/paper/1909.03403/) (2019)
- [Towards Understanding Fast Adversarial Training](https://mlsanity.com/paper/2006.03089/) (2020)
- [The Domain Shift Problem of Medical Image Segmentation and Vendor-Adaptation by Unet-GAN](https://mlsanity.com/paper/1910.13681/) (2019)
- [Neural Style Transfer Improves 3D Cardiovascular MR Image Segmentation on Inconsistent Data](https://mlsanity.com/paper/1909.09716/) (2019)
- [Unsupervised Multi-modal Style Transfer for Cardiac MR Segmentation](https://mlsanity.com/paper/1908.07344/) (2019)
- [Remove Appearance Shift for Ultrasound Image Segmentation via Fast and Universal Style Transfer](https://mlsanity.com/paper/2002.05844/) (2020)
- [Style is a Distribution of Features](https://mlsanity.com/paper/2007.13010/) (2020)

## Built on by (most cited first, 6 of 6)

- [Online Reflective Learning for Robust Medical Image Segmentation](https://mlsanity.com/paper/2207.00476/) (2022)
- [Supervision by Denoising for Medical Image Segmentation](https://mlsanity.com/paper/2202.02952/) (2022)
- [Progressive Curriculum Learning with Scale-Enhanced U-Net for Continuous Airway Segmentation](https://mlsanity.com/paper/2410.18456/) (2024)
- [FoundBioNet: A Foundation-Based Model for IDH Genotyping of Glioma from Multi-Parametric MRI](https://mlsanity.com/paper/2508.06756/) (2025)
- [Progressive Learning with Anatomical Priors for Reliable Left Atrial Scar Segmentation from Late Gadolinium Enhancement MRI](https://mlsanity.com/paper/2603.26186/) (2026)
- [MedLoc-R1: Performance-Aware Curriculum Reward Scheduling for GRPO-Based Medical Visual Grounding](https://mlsanity.com/paper/2603.28120/) (2026)

## Abstract (arXiv)

The performance of deep segmentation models often degrades due to distribution shifts in image intensities between the training and test data sets. This is particularly pronounced in multi-centre studies involving data acquired using multi-vendor scanners, with variations in acquisition protocols. It is challenging to address this degradation because the shift is often not known textita priori and hence difficult to model. We propose a novel framework to ensure robust segmentation in the presence of such distribution shifts. Our contribution is three-fold. First, inspired by the spirit of curriculum learning, we design a novel style curriculum to train the segmentation models using an easy-to-hard mode. A style transfer model with style fusion is employed to generate the curriculum samples. Gradually focusing on complex and adversarial style samples can significantly boost the robustness of the models. Second, instead of subjectively defining the curriculum complexity, we adopt an automated gradient manipulation method to control the hard and adversarial sample generation process. Third, we propose the Local Gradient Sign strategy to aggregate the gradient locally and stabilise training during gradient manipulation. The proposed framework can generalise to unknown distribution without using any target data. Extensive experiments on the public M&Ms Challenge dataset demonstrate that our proposed framework can generalise deep models well to unknown distributions and achieve significant improvements in segmentation accuracy.