A knowledge-based method for detecting network-induced shape artifacts in synthetic images

Rucha Deshpande, Miguel Lago, Adarsh Subbaswamy, Seyed Kahaki, Jana G Delfino, Aldo Badano, Ghada Zamzmi
Proceedings of The 8th International Conference on Medical Imaging with Deep Learning, PMLR 301:346-365, 2026.

Abstract

The adoption of synthetic medical images for training or testing without thorough quality assessment risks introducing artifacts and unrealistic features that can mislead machine learning models and compromise clinical utility. This work introduces a novel knowledge-based method for detecting network-induced shape artifacts in synthetic images. The method can identify anatomically unrealistic images, detect shape artifacts irrespective of the generative model, and offer interpretability through its knowledge-driven design. We validated the method using two synthetic mammography datasets and demonstrated its effectiveness in flagging images with network-induced artifacts. A reader study further confirmed these findings and showed that the most anomalous images identified by the method were also flagged by human readers. This method provides a step toward the responsible use of synthetic data by ensuring synthetic images align with realistic morphological and anatomical constraints.

Cite this Paper

BibTeX
@InProceedings{pmlr-v301-deshpande26a, title = {A knowledge-based method for detecting network-induced shape artifacts in synthetic images}, author = {Deshpande, Rucha and Lago, Miguel and Subbaswamy, Adarsh and Kahaki, Seyed and Delfino, Jana G and Badano, Aldo and Zamzmi, Ghada}, booktitle = {Proceedings of The 8th International Conference on Medical Imaging with Deep Learning}, pages = {346--365}, 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/deshpande26a/deshpande26a.pdf}, url = {https://proceedings.mlr.press/v301/deshpande26a.html}, abstract = {The adoption of synthetic medical images for training or testing without thorough quality assessment risks introducing artifacts and unrealistic features that can mislead machine learning models and compromise clinical utility. This work introduces a novel knowledge-based method for detecting network-induced shape artifacts in synthetic images. The method can identify anatomically unrealistic images, detect shape artifacts irrespective of the generative model, and offer interpretability through its knowledge-driven design. We validated the method using two synthetic mammography datasets and demonstrated its effectiveness in flagging images with network-induced artifacts. A reader study further confirmed these findings and showed that the most anomalous images identified by the method were also flagged by human readers. This method provides a step toward the responsible use of synthetic data by ensuring synthetic images align with realistic morphological and anatomical constraints.} }
Endnote
%0 Conference Paper %T A knowledge-based method for detecting network-induced shape artifacts in synthetic images %A Rucha Deshpande %A Miguel Lago %A Adarsh Subbaswamy %A Seyed Kahaki %A Jana G Delfino %A Aldo Badano %A Ghada Zamzmi %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-deshpande26a %I PMLR %P 346--365 %U https://proceedings.mlr.press/v301/deshpande26a.html %V 301 %X The adoption of synthetic medical images for training or testing without thorough quality assessment risks introducing artifacts and unrealistic features that can mislead machine learning models and compromise clinical utility. This work introduces a novel knowledge-based method for detecting network-induced shape artifacts in synthetic images. The method can identify anatomically unrealistic images, detect shape artifacts irrespective of the generative model, and offer interpretability through its knowledge-driven design. We validated the method using two synthetic mammography datasets and demonstrated its effectiveness in flagging images with network-induced artifacts. A reader study further confirmed these findings and showed that the most anomalous images identified by the method were also flagged by human readers. This method provides a step toward the responsible use of synthetic data by ensuring synthetic images align with realistic morphological and anatomical constraints.
APA
Deshpande, R., Lago, M., Subbaswamy, A., Kahaki, S., Delfino, J.G., Badano, A. & Zamzmi, G.. (2026). A knowledge-based method for detecting network-induced shape artifacts in synthetic images. Proceedings of The 8th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 301:346-365 Available from https://proceedings.mlr.press/v301/deshpande26a.html.

Related Material