Unsupervised Patch-GAN with Targeted Patch Ranking for Fine-Grained Novelty Detection in Medical Imaging

Jingkun Chen, Guang Yang, Xiao Zhang, Jingchao Peng, Tianlu Zhang, Jianguo Zhang, Jungong Han, Vicente Grau
Proceedings of The First AAAI Bridge Program on AI for Medicine and Healthcare, PMLR 281:35-41, 2025.

Abstract

Detecting novel anomalies in medical imaging is challenging due to the limited availability of labeled data for rare abnormalities, which often display high variability and subtlety. This challenge is further compounded when small abnormal regions are embedded within larger normal areas, as whole-image predictions frequently overlook these subtle deviations. To address these issues, we propose an unsupervised Patch-GAN framework designed to detect and localize anomalies by capturing both local detail and global structure. Our framework first reconstructs masked images to learn fine-grained, normal-specific features, allowing for enhanced sensitivity to minor deviations from normality. By dividing these reconstructed images into patches and assessing the authenticity of each patch, our approach identifies anomalies at a more granular level, overcoming the limitations of whole-image evaluation. Additionally, a patch-ranking mechanism prioritizes regions with higher abnormal scores, reinforcing the alignment between local patch discrepancies and the global image context. Experimental results on the ISIC 2016 skin lesion and BraTS 2019 brain tumor datasets validate our framework’s effectiveness, achieving AUCs of 95.79% and 96.05%, respectively, and outperforming three state-of-the-art baselines.

Cite this Paper

BibTeX
@InProceedings{pmlr-v281-chen25a, title = {Unsupervised Patch-GAN with Targeted Patch Ranking for Fine-Grained Novelty Detection in Medical Imaging}, author = {Chen, Jingkun and Yang, Guang and Zhang, Xiao and Peng, Jingchao and Zhang, Tianlu and Zhang, Jianguo and Han, Jungong and Grau, Vicente}, booktitle = {Proceedings of The First AAAI Bridge Program on AI for Medicine and Healthcare}, pages = {35--41}, year = {2025}, editor = {Wu, Junde and Zhu, Jiayuan and Xu, Min and Jin, Yueming}, volume = {281}, series = {Proceedings of Machine Learning Research}, month = {25 Feb}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v281/main/assets/chen25a/chen25a.pdf}, url = {https://proceedings.mlr.press/v281/chen25a.html}, abstract = {Detecting novel anomalies in medical imaging is challenging due to the limited availability of labeled data for rare abnormalities, which often display high variability and subtlety. This challenge is further compounded when small abnormal regions are embedded within larger normal areas, as whole-image predictions frequently overlook these subtle deviations. To address these issues, we propose an unsupervised Patch-GAN framework designed to detect and localize anomalies by capturing both local detail and global structure. Our framework first reconstructs masked images to learn fine-grained, normal-specific features, allowing for enhanced sensitivity to minor deviations from normality. By dividing these reconstructed images into patches and assessing the authenticity of each patch, our approach identifies anomalies at a more granular level, overcoming the limitations of whole-image evaluation. Additionally, a patch-ranking mechanism prioritizes regions with higher abnormal scores, reinforcing the alignment between local patch discrepancies and the global image context. Experimental results on the ISIC 2016 skin lesion and BraTS 2019 brain tumor datasets validate our framework’s effectiveness, achieving AUCs of 95.79% and 96.05%, respectively, and outperforming three state-of-the-art baselines.} }
Endnote
%0 Conference Paper %T Unsupervised Patch-GAN with Targeted Patch Ranking for Fine-Grained Novelty Detection in Medical Imaging %A Jingkun Chen %A Guang Yang %A Xiao Zhang %A Jingchao Peng %A Tianlu Zhang %A Jianguo Zhang %A Jungong Han %A Vicente Grau %B Proceedings of The First AAAI Bridge Program on AI for Medicine and Healthcare %C Proceedings of Machine Learning Research %D 2025 %E Junde Wu %E Jiayuan Zhu %E Min Xu %E Yueming Jin %F pmlr-v281-chen25a %I PMLR %P 35--41 %U https://proceedings.mlr.press/v281/chen25a.html %V 281 %X Detecting novel anomalies in medical imaging is challenging due to the limited availability of labeled data for rare abnormalities, which often display high variability and subtlety. This challenge is further compounded when small abnormal regions are embedded within larger normal areas, as whole-image predictions frequently overlook these subtle deviations. To address these issues, we propose an unsupervised Patch-GAN framework designed to detect and localize anomalies by capturing both local detail and global structure. Our framework first reconstructs masked images to learn fine-grained, normal-specific features, allowing for enhanced sensitivity to minor deviations from normality. By dividing these reconstructed images into patches and assessing the authenticity of each patch, our approach identifies anomalies at a more granular level, overcoming the limitations of whole-image evaluation. Additionally, a patch-ranking mechanism prioritizes regions with higher abnormal scores, reinforcing the alignment between local patch discrepancies and the global image context. Experimental results on the ISIC 2016 skin lesion and BraTS 2019 brain tumor datasets validate our framework’s effectiveness, achieving AUCs of 95.79% and 96.05%, respectively, and outperforming three state-of-the-art baselines.
APA
Chen, J., Yang, G., Zhang, X., Peng, J., Zhang, T., Zhang, J., Han, J. & Grau, V.. (2025). Unsupervised Patch-GAN with Targeted Patch Ranking for Fine-Grained Novelty Detection in Medical Imaging. Proceedings of The First AAAI Bridge Program on AI for Medicine and Healthcare, in Proceedings of Machine Learning Research 281:35-41 Available from https://proceedings.mlr.press/v281/chen25a.html.

Related Material