Revealing and Reducing Morphological Biases Using Implicit Neural Representations for Medical Image Registration

Sofija Engelson, Bennet Kahrs, Timo Kepp, Julia Andresen, Heinz Handels, Jan Ehrhardt
Proceedings of The 9th International Conference on Medical Imaging with Deep Learning, PMLR 315:4026-4041, 2026.

Abstract

Deep learning has enhanced medical image analysis, yet models trained on imbalanced or non-representative populations often exhibit systematic biases, which can lead to substantial performance disparities across patient subgroups. Addressing these disparities is essential to ensure fair and reliable model deployment in clinical practice. Particularly in medical imaging, population-level biases can oftentimes be attributed to morphological rather than intensity differences, such as sex-related differences in organ volume. Given that morphological biases in neuroimaging data spuriously correlate with the disease label, we show, that bias detection based on general foundation model features (e.g., CLIP and BiomedCLIP) insufficiently captures morphological biases. Therefore, we introduce a bias detection and mitigation pipeline that performs subgroup discovery on deformation representations from a generalizable implicit neural representation (INR). This proof-of-concept study indicates improved performance when using deformation representations instead of general image features for bias detection. Furthermore, our results show that re-balancing the training dataset using the identified subgroups, complemented by INR-generated samples for augmentation, helps to mitigate the bias effect.

Cite this Paper

BibTeX
@InProceedings{pmlr-v315-engelson26a, title = {Revealing and Reducing Morphological Biases Using Implicit Neural Representations for Medical Image Registration}, author = {Engelson, Sofija and Kahrs, Bennet and Kepp, Timo and Andresen, Julia and Handels, Heinz and Ehrhardt, Jan}, booktitle = {Proceedings of The 9th International Conference on Medical Imaging with Deep Learning}, pages = {4026--4041}, 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/engelson26a/engelson26a.pdf}, url = {https://proceedings.mlr.press/v315/engelson26a.html}, abstract = {Deep learning has enhanced medical image analysis, yet models trained on imbalanced or non-representative populations often exhibit systematic biases, which can lead to substantial performance disparities across patient subgroups. Addressing these disparities is essential to ensure fair and reliable model deployment in clinical practice. Particularly in medical imaging, population-level biases can oftentimes be attributed to morphological rather than intensity differences, such as sex-related differences in organ volume. Given that morphological biases in neuroimaging data spuriously correlate with the disease label, we show, that bias detection based on general foundation model features (e.g., CLIP and BiomedCLIP) insufficiently captures morphological biases. Therefore, we introduce a bias detection and mitigation pipeline that performs subgroup discovery on deformation representations from a generalizable implicit neural representation (INR). This proof-of-concept study indicates improved performance when using deformation representations instead of general image features for bias detection. Furthermore, our results show that re-balancing the training dataset using the identified subgroups, complemented by INR-generated samples for augmentation, helps to mitigate the bias effect.} }
Endnote
%0 Conference Paper %T Revealing and Reducing Morphological Biases Using Implicit Neural Representations for Medical Image Registration %A Sofija Engelson %A Bennet Kahrs %A Timo Kepp %A Julia Andresen %A Heinz Handels %A Jan Ehrhardt %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-engelson26a %I PMLR %P 4026--4041 %U https://proceedings.mlr.press/v315/engelson26a.html %V 315 %X Deep learning has enhanced medical image analysis, yet models trained on imbalanced or non-representative populations often exhibit systematic biases, which can lead to substantial performance disparities across patient subgroups. Addressing these disparities is essential to ensure fair and reliable model deployment in clinical practice. Particularly in medical imaging, population-level biases can oftentimes be attributed to morphological rather than intensity differences, such as sex-related differences in organ volume. Given that morphological biases in neuroimaging data spuriously correlate with the disease label, we show, that bias detection based on general foundation model features (e.g., CLIP and BiomedCLIP) insufficiently captures morphological biases. Therefore, we introduce a bias detection and mitigation pipeline that performs subgroup discovery on deformation representations from a generalizable implicit neural representation (INR). This proof-of-concept study indicates improved performance when using deformation representations instead of general image features for bias detection. Furthermore, our results show that re-balancing the training dataset using the identified subgroups, complemented by INR-generated samples for augmentation, helps to mitigate the bias effect.
APA
Engelson, S., Kahrs, B., Kepp, T., Andresen, J., Handels, H. & Ehrhardt, J.. (2026). Revealing and Reducing Morphological Biases Using Implicit Neural Representations for Medical Image Registration. Proceedings of The 9th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 315:4026-4041 Available from https://proceedings.mlr.press/v315/engelson26a.html.

Related Material