Brain Tumor Growth Inversion via Differentiable Neural Surrogates

Jonas Weidner, Lucas Zimmer, Ivan Ezhov, Michal Balcerak, Björn Menze, Daniel Rückert, Benedikt Wiestler
Proceedings of The 9th International Conference on Medical Imaging with Deep Learning, PMLR 315:781-800, 2026.

Abstract

Personalizing biophysical brain tumor models to individual patients is computationally expensive due to the need for numerous iterative evaluations of slow numerical solvers to identify optimal patient-specific parameters. We address this by introducing a differentiable neural surrogate that replaces the traditional forward model. Unlike the original solver, this surrogate is fully differentiable, allowing us to solve the inverse problem using highly efficient gradient-based optimization. This approach ensures that the solution learns the biophysical constraints of tumor growth while accelerating the process by orders of magnitude. In a 3D brain tumor growth setting, our framework achieves accuracy competitive with classical optimization while reducing runtime from days to seconds. Crucially, we demonstrate that our method, though trained on synthetic data, generalizes effectively to real patient scans. These findings establish differentiable surrogates as a powerful tool for accelerating scientific machine learning in medical imaging and beyond.

Cite this Paper

BibTeX
@InProceedings{pmlr-v315-weidner26a, title = {Brain Tumor Growth Inversion via Differentiable Neural Surrogates}, author = {Weidner, Jonas and Zimmer, Lucas and Ezhov, Ivan and Balcerak, Michal and Menze, Bj{\"o}rn and R{\"u}ckert, Daniel and Wiestler, Benedikt}, booktitle = {Proceedings of The 9th International Conference on Medical Imaging with Deep Learning}, pages = {781--800}, 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/weidner26a/weidner26a.pdf}, url = {https://proceedings.mlr.press/v315/weidner26a.html}, abstract = {Personalizing biophysical brain tumor models to individual patients is computationally expensive due to the need for numerous iterative evaluations of slow numerical solvers to identify optimal patient-specific parameters. We address this by introducing a differentiable neural surrogate that replaces the traditional forward model. Unlike the original solver, this surrogate is fully differentiable, allowing us to solve the inverse problem using highly efficient gradient-based optimization. This approach ensures that the solution learns the biophysical constraints of tumor growth while accelerating the process by orders of magnitude. In a 3D brain tumor growth setting, our framework achieves accuracy competitive with classical optimization while reducing runtime from days to seconds. Crucially, we demonstrate that our method, though trained on synthetic data, generalizes effectively to real patient scans. These findings establish differentiable surrogates as a powerful tool for accelerating scientific machine learning in medical imaging and beyond.} }
Endnote
%0 Conference Paper %T Brain Tumor Growth Inversion via Differentiable Neural Surrogates %A Jonas Weidner %A Lucas Zimmer %A Ivan Ezhov %A Michal Balcerak %A Björn Menze %A Daniel Rückert %A Benedikt Wiestler %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-weidner26a %I PMLR %P 781--800 %U https://proceedings.mlr.press/v315/weidner26a.html %V 315 %X Personalizing biophysical brain tumor models to individual patients is computationally expensive due to the need for numerous iterative evaluations of slow numerical solvers to identify optimal patient-specific parameters. We address this by introducing a differentiable neural surrogate that replaces the traditional forward model. Unlike the original solver, this surrogate is fully differentiable, allowing us to solve the inverse problem using highly efficient gradient-based optimization. This approach ensures that the solution learns the biophysical constraints of tumor growth while accelerating the process by orders of magnitude. In a 3D brain tumor growth setting, our framework achieves accuracy competitive with classical optimization while reducing runtime from days to seconds. Crucially, we demonstrate that our method, though trained on synthetic data, generalizes effectively to real patient scans. These findings establish differentiable surrogates as a powerful tool for accelerating scientific machine learning in medical imaging and beyond.
APA
Weidner, J., Zimmer, L., Ezhov, I., Balcerak, M., Menze, B., Rückert, D. & Wiestler, B.. (2026). Brain Tumor Growth Inversion via Differentiable Neural Surrogates. Proceedings of The 9th International Conference on Medical Imaging with Deep Learning, in Proceedings of Machine Learning Research 315:781-800 Available from https://proceedings.mlr.press/v315/weidner26a.html.

Related Material