arXiv subjects

Jelmer M. Wolterink

Publications and source records attributed to Jelmer M. Wolterink.

At least 19 recordsLinked to original sources

arXiv2026

When Does Equivariance Help? Canonical Alignment in Neural Fluid Surrogates

Neural surrogates can accelerate computational fluid dynamics (CFD) simulations by orders of magnitude, but practical deployment in engineering and healthcare applications requires architectures that scale to high-resolution meshes and learn effectively from limited data. Explicit equivariance offers a principled inductive bias, yet its accuracy benefits may depend on the prediction task and the distribution of anatomical orientations. We investigate this dependence across three hemodynamic benchmarks with different degrees of natural canonical alignment. To support this study, we introduce the Anchored-Branched Geometric Algebra Transformer (AB-GATr), an $E(3)$-equivariant surrogate that efficiently predicts coupled surface and volume quantities. Across these benchmarks, AB-GATr consistently outperforms the evaluated non-equivariant models, including variants trained with rotational augmentation, while achieving accuracy competitive with $E(3)$-equivariant LaB-GATr at substantially lower training cost. In comparison, rotational augmentation provides inconsistent benefits across architectures and can reduce accuracy. A controlled experiment on ShapeNet-Car shows that strong canonical alignment can favor non-equivariant models, but their accuracy generally deteriorates as training orientations broaden and can decline sharply under broader test rotations. We further investigate these patterns using extended symmetry-breaking diagnostics and probes of the predictive information associated with canonical alignment across all benchmarks. Together, these results support explicit equivariance for the evaluated hemodynamic tasks with natural orientation variation, while showing that its accuracy benefits depend on the task and orientation distribution.

cs.LG↗

arXiv2026

Towards Brain MRI Foundation Models for the Clinic: Findings from the FOMO25 Challenge

Clinical deployment of automated brain MRI analysis faces a fundamental challenge: clinical data is heterogeneous and noisy, and high-quality labels are prohibitively costly to obtain. Self-supervised learning (SSL) can address this by leveraging the vast amounts of unlabeled data produced in clinical workflows to train robust \textit{foundation models} that adapt out-of-domain with minimal supervision. However, the development of foundation models for brain MRI has been limited by small pretraining datasets and in-domain benchmarking focused on high-quality, research-grade data. To address this gap, we organized the FOMO25 challenge as a satellite event at MICCAI 2025. FOMO25 provided participants with a large pretraining dataset, FOMO60K, and evaluated models on data sourced directly from clinical workflows in few-shot and out-of-domain settings. Tasks covered infarct classification, meningioma segmentation, and brain age regression, and considered both models trained on FOMO60K (method track) and any data (open track). Nineteen foundation models from sixteen teams were evaluated using a standardized containerized pipeline. Results show that (a) self-supervised pretraining improves generalization on clinical data under domain shift, with the strongest models trained \textit{out-of-domain} surpassing supervised baselines trained \textit{in-domain}. (b) No single pretraining objective benefits all tasks: MAE favors segmentation, hybrid reconstruction-contrastive objectives favor classification, and (c) strong performance was achieved by small pretrained models, and improvements from scaling model size and training duration did not yield reliable benefits.

cs.CV↗

arXiv2026

SegReg: Latent Space Regularization for Improved Medical Image Segmentation

Medical image segmentation models are typically optimised with voxel-wise losses that constrain predictions only in the output space. This leaves latent feature representations largely unconstrained, potentially limiting generalisation. We propose {SegReg}, a latent-space regularisation framework that operates on feature maps of U-Net models to encourage structured embeddings while remaining fully compatible with standard segmentation losses. Integrated with the nnU-Net framework, we evaluate SegReg on prostate, cardiac, and hippocampus segmentation and demonstrate consistent improvements in domain generalisation. Furthermore, we show that explicit latent regularisation improves continual learning by reducing task drift and enhancing forward transfer across sequential tasks without adding memory or any extra parameters. These results highlight latent-space regularisation as a practical approach for building more generalisable and continual-learning-ready models.

eess.IV↗

arXiv2026

Physics-informed graph neural networks for flow field estimation in carotid arteries

Hemodynamic quantities are valuable biomedical risk factors for cardiovascular pathology such as atherosclerosis. Non-invasive, in-vivo measurement of these quantities can only be performed using a select number of modalities that are not widely available, such as 4D flow magnetic resonance imaging (MRI). In this work, we create a surrogate model for hemodynamic flow field estimation, powered by machine learning. We train graph neural networks that include priors about the underlying symmetries and physics, limiting the amount of data required for training. This allows us to train the model using moderately-sized, in-vivo 4D flow MRI datasets, instead of large in-silico datasets obtained by computational fluid dynamics (CFD), as is the current standard. We create an efficient, equivariant neural network by combining the popular PointNet++ architecture with group-steerable layers. To incorporate the physics-informed priors, we derive an efficient discretisation scheme for the involved differential operators. We perform extensive experiments in carotid arteries and show that our model can accurately estimate low-noise hemodynamic flow fields in the carotid artery. Moreover, we show how the learned relation between geometry and hemodynamic quantities transfers to 3D vascular models obtained using a different imaging modality than the training data. This shows that physics-informed graph neural networks can be trained using 4D flow MRI data to estimate blood flow in unseen carotid artery geometries.

q-bio.QM↗

arXiv2026

Knowledge Distillation for Continual Learning of Biomedical Neural Fields

Neural fields are increasingly used as a light-weight, continuous, and differentiable signal representation in (bio)medical imaging. However, unlike discrete signal representations such as voxel grids, neural fields cannot be easily extended. As neural fields are, in essence, neural networks, prior signals represented in a neural field will degrade when the model is presented with new data due to catastrophic forgetting. This work examines the extent to which different neural field approaches suffer from catastrophic forgetting and proposes a strategy to mitigate this issue. We consider the scenario in which data becomes available incrementally, with only the most recent data available for neural field fitting. In a series of experiments on cardiac cine MRI data, we demonstrate how knowledge distillation mitigates catastrophic forgetting when the spatiotemporal domain is enlarged or the dimensionality of the represented signal is increased. We find that the amount of catastrophic forgetting depends, to a large extent, on the neural fields model used, and that distillation could enable continual learning in neural fields.

eess.IV↗

arXiv2025

A deep learning model to reduce agent dose for contrast-enhanced MRI of the cerebellopontine angle cistern

Objectives: To evaluate a deep learning (DL) model for reducing the agent dose of contrast-enhanced T1-weighted MRI (T1ce) of the cerebellopontine angle (CPA) cistern. Materials and methods: In this multi-center retrospective study, T1 and T1ce of vestibular schwannoma (VS) patients were used to simulate low-dose T1ce with varying reductions of contrast agent dose. DL models were trained to restore standard-dose T1ce from the low-dose simulation. The image quality and segmentation performance of the DL-restored T1ce were evaluated. A head and neck radiologist was asked to rate DL-restored images in multiple aspects, including image quality and diagnostic characterization. Results: 203 MRI studies from 72 VS patients (mean age, 58.51 \pm 14.73, 39 men) were evaluated. As the input dose increased, the structural similarity index measure of the restored T1ce increased from 0.639 \pm 0.113 to 0.993 \pm 0.009, and the peak signal-to-noise ratio increased from 21.6 \pm 3.73 dB to 41.4 \pm 4.84 dB. At 10% input dose, using DL-restored T1ce for segmentation improved the Dice from 0.673 to 0.734, the 95% Hausdorff distance from 2.38 mm to 2.07 mm, and the average surface distance from 1.00 mm to 0.59 mm. Both DL-restored T1ce from 10% and 30% input doses showed excellent images, with the latter being considered more informative. Conclusion: The DL model improved the image quality of low-dose MRI of the CPA cistern, which makes lesion detection and diagnostic characterization possible with 10% - 30% of the standard dose.

cs.CV↗

arXiv2025

Consistent View Alignment Improves Foundation Models for 3D Medical Image Segmentation

Many recent approaches in representation learning implicitly assume that uncorrelated views of a data point are sufficient to learn meaningful representations for various downstream tasks. In this work, we challenge this assumption and demonstrate that meaningful structure in the latent space does not emerge naturally. Instead, it must be explicitly induced. We propose a method that aligns representations from different views of the data to align complementary information without inducing false positives. Our experiments show that our proposed self-supervised learning method, Consistent View Alignment, improves performance for downstream tasks, highlighting the critical role of structured view alignment in learning effective representations. Our method achieved first and second place in the MICCAI 2025 SSL3D challenge when using a Primus vision transformer and ResEnc convolutional neural network, respectively. The code and pretrained model weights are released at https://github.com/Tenbatsu24/LatentCampus.

cs.CV↗

arXiv2025

Active Learning for Deep Learning-Based Hemodynamic Parameter Estimation

Hemodynamic parameters such as pressure and wall shear stress play an important role in diagnosis, prognosis, and treatment planning in cardiovascular diseases. These parameters can be accurately computed using computational fluid dynamics (CFD), but CFD is computationally intensive. Hence, deep learning methods have been adopted as a surrogate to rapidly estimate CFD outcomes. A drawback of such data-driven models is the need for time-consuming reference CFD simulations for training. In this work, we introduce an active learning framework to reduce the number of CFD simulations required for the training of surrogate models, lowering the barriers to their deployment in new applications. We propose three distinct querying strategies to determine for which unlabeled samples CFD simulations should be obtained. These querying strategies are based on geometrical variance, ensemble uncertainty, and adherence to the physics governing fluid dynamics. We benchmark these methods on velocity field estimation in synthetic coronary artery bifurcations and find that they allow for substantial reductions in annotation cost. Notably, we find that our strategies reduce the number of samples required by up to 50% and make the trained models more robust to difficult cases. Our results show that active learning is a feasible strategy to increase the potential of deep learning-based CFD surrogates.

cs.CV↗

arXiv2025

GReAT: leveraging geometric artery data to improve wall shear stress assessment

Leveraging big data for patient care is promising in many medical fields such as cardiovascular health. For example, hemodynamic biomarkers like wall shear stress could be assessed from patient-specific medical images via machine learning algorithms, bypassing the need for time-intensive computational fluid simulation. However, it is extremely challenging to amass large-enough datasets to effectively train such models. We could address this data scarcity by means of self-supervised pre-training and foundations models given large datasets of geometric artery models. In the context of coronary arteries, leveraging learned representations to improve hemodynamic biomarker assessment has not yet been well studied. In this work, we address this gap by investigating whether a large dataset (8449 shapes) consisting of geometric models of 3D blood vessels can benefit wall shear stress assessment in coronary artery models from a small-scale clinical trial (49 patients). We create a self-supervised target for the 3D blood vessels by computing the heat kernel signature, a quantity obtained via Laplacian eigenvectors, which captures the very essence of the shapes. We show how geometric representations learned from this datasets can boost segmentation of coronary arteries into regions of low, mid and high (time-averaged) wall shear stress even when trained on limited data.

cs.CV↗

arXiv2025

Deep vectorised operators for pulsatile hemodynamics estimation in coronary arteries from a steady-state prior

Cardiovascular hemodynamic fields provide valuable medical decision markers for coronary artery disease. Computational fluid dynamics (CFD) is the gold standard for accurate, non-invasive evaluation of these quantities in silico. In this work, we propose a time-efficient surrogate model, powered by machine learning, for the estimation of pulsatile hemodynamics based on steady-state priors. We introduce deep vectorised operators, a modelling framework for discretisation-independent learning on infinite-dimensional function spaces. The underlying neural architecture is a neural field conditioned on hemodynamic boundary conditions. Importantly, we show how relaxing the requirement of point-wise action to permutation-equivariance leads to a family of models that can be parametrised by message passing and self-attention layers. We evaluate our approach on a dataset of 74 stenotic coronary arteries extracted from coronary computed tomography angiography (CCTA) with patient-specific pulsatile CFD simulations as ground truth. We show that our model produces accurate estimates of the pulsatile velocity and pressure (approximation disparity 0.368 $\pm$ 0.079) while being agnostic ($p < 0.05$ in a one-way ANOVA test) to re-sampling of the source domain, i.e. discretisation-independent. This shows that deep vectorised operators are a powerful modelling tool for cardiovascular hemodynamics estimation in coronary arteries and beyond.

q-bio.QM↗

arXiv2025

Beyond Pixels: Medical Image Quality Assessment with Implicit Neural Representations

Artifacts pose a significant challenge in medical imaging, impacting diagnostic accuracy and downstream analysis. While image-based approaches for detecting artifacts can be effective, they often rely on preprocessing methods that can lead to information loss and high-memory-demand medical images, thereby limiting the scalability of classification models. In this work, we propose the use of implicit neural representations (INRs) for image quality assessment. INRs provide a compact and continuous representation of medical images, naturally handling variations in resolution and image size while reducing memory overhead. We develop deep weight space networks, graph neural networks, and relational attention transformers that operate on INRs to achieve image quality assessment. Our method is evaluated on the ACDC dataset with synthetically generated artifact patterns, demonstrating its effectiveness in assessing image quality while achieving similar performance with fewer parameters.

eess.IV↗

arXiv2025

Wall Shear Stress Estimation in Abdominal Aortic Aneurysms: Towards Generalisable Neural Surrogate Models

Abdominal aortic aneurysms (AAAs) are pathologic dilatations of the abdominal aorta posing a high fatality risk upon rupture. Studying AAA progression and rupture risk often involves in-silico blood flow modelling with computational fluid dynamics (CFD) and extraction of hemodynamic factors like time-averaged wall shear stress (TAWSS) or oscillatory shear index (OSI). However, CFD simulations are known to be computationally demanding. Hence, in recent years, geometric deep learning methods, operating directly on 3D shapes, have been proposed as compelling surrogates, estimating hemodynamic parameters in just a few seconds. In this work, we propose a geometric deep learning approach to estimating hemodynamics in AAA patients, and study its generalisability to common factors of real-world variation. We propose an E(3)-equivariant deep learning model utilising novel robust geometrical descriptors and projective geometric algebra. Our model is trained to estimate transient WSS using a dataset of CT scans of 100 AAA patients, from which lumen geometries are extracted and reference CFD simulations with varying boundary conditions are obtained. Results show that the model generalizes well within the distribution, as well as to the external test set. Moreover, the model can accurately estimate hemodynamics across geometry remodelling and changes in boundary conditions. Furthermore, we find that a trained model can be applied to different artery tree topologies, where new and unseen branches are added during inference. Finally, we find that the model is to a large extent agnostic to mesh resolution. These results show the accuracy and generalisation of the proposed model, and highlight its potential to contribute to hemodynamic parameter estimation in clinical practice.

cs.CV↗

arXiv2025

Geometric deep learning for local growth prediction on abdominal aortic aneurysm surfaces

Abdominal aortic aneurysms (AAAs) are progressive focal dilatations of the abdominal aorta. AAAs may rupture, with a survival rate of only 20\%. Current clinical guidelines recommend elective surgical repair when the maximum AAA diameter exceeds 55 mm in men or 50 mm in women. Patients that do not meet these criteria are periodically monitored, with surveillance intervals based on the maximum AAA diameter. However, this diameter does not take into account the complex relation between the 3D AAA shape and its growth, making standardized intervals potentially unfit. Personalized AAA growth predictions could improve monitoring strategies. We propose to use an SE(3)-symmetric transformer model to predict AAA growth directly on the vascular model surface enriched with local, multi-physical features. In contrast to other works which have parameterized the AAA shape, this representation preserves the vascular surface's anatomical structure and geometric fidelity. We train our model using a longitudinal dataset of 113 computed tomography angiography (CTA) scans of 24 AAA patients at irregularly sampled intervals. After training, our model predicts AAA growth to the next scan moment with a median diameter error of 1.18 mm. We further demonstrate our model's utility to identify whether a patient will become eligible for elective repair within two years (acc = 0.93). Finally, we evaluate our model's generalization on an external validation set consisting of 25 CTAs from 7 AAA patients from a different hospital. Our results show that local directional AAA growth prediction from the vascular surface is feasible and may contribute to personalized surveillance strategies.

cs.CV↗

arXiv2025

Data-Agnostic Augmentations for Unknown Variations: Out-of-Distribution Generalisation in MRI Segmentation

Medical image segmentation models are often trained on curated datasets, leading to performance degradation when deployed in real-world clinical settings due to mismatches between training and test distributions. While data augmentation techniques are widely used to address these challenges, traditional visually consistent augmentation strategies lack the robustness needed for diverse real-world scenarios. In this work, we systematically evaluate alternative augmentation strategies, focusing on MixUp and Auxiliary Fourier Augmentation. These methods mitigate the effects of multiple variations without explicitly targeting specific sources of distribution shifts. We demonstrate how these techniques significantly improve out-of-distribution generalization and robustness to imaging variations across a wide range of transformations in cardiac cine MRI and prostate MRI segmentation. We quantitatively find that these augmentation methods enhance learned feature representations by promoting separability and compactness. Additionally, we highlight how their integration into nnU-Net training pipelines provides an easy-to-implement, effective solution for enhancing the reliability of medical segmentation models in real-world applications.

cs.CV↗

arXiv2025

Steerable Anatomical Shape Synthesis with Implicit Neural Representations

Generative modeling of anatomical structures plays a crucial role in virtual imaging trials, which allow researchers to perform studies without the costs and constraints inherent to in vivo and phantom studies. For clinical relevance, generative models should allow targeted control to simulate specific patient populations rather than relying on purely random sampling. In this work, we propose a steerable generative model based on implicit neural representations. Implicit neural representations naturally support topology changes, making them well-suited for anatomical structures with varying topology, such as the thyroid. Our model learns a disentangled latent representation, enabling fine-grained control over shape variations. Evaluation includes reconstruction accuracy and anatomical plausibility. Our results demonstrate that the proposed model achieves high-quality shape generation while enabling targeted anatomical modifications.

cs.CV↗

arXiv2025

World of Forms: Deformable Geometric Templates for One-Shot Surface Meshing in Coronary CT Angiography

Deep learning-based medical image segmentation and surface mesh generation typically involve a sequential pipeline from image to segmentation to meshes, often requiring large training datasets while making limited use of prior geometric knowledge. This may lead to topological inconsistencies and suboptimal performance in low-data regimes. To address these challenges, we propose a data-efficient deep learning method for direct 3D anatomical object surface meshing using geometric priors. Our approach employs a multi-resolution graph neural network that operates on a prior geometric template which is deformed to fit object boundaries of interest. We show how different templates may be used for the different surface meshing targets, and introduce a novel masked autoencoder pretraining strategy for 3D spherical data. The proposed method outperforms nnUNet in a one-shot setting for segmentation of the pericardium, left ventricle (LV) cavity and the LV myocardium. Similarly, the method outperforms other lumen segmentation operating on multi-planar reformatted images. Results further indicate that mesh quality is on par with or improves upon marching cubes post-processing of voxel mask predictions, while remaining flexible in the choice of mesh triangulation prior, thus paving the way for more accurate and topologically consistent 3D medical object surface meshing.

eess.IV↗

arXiv2025

Waveform-Specific Performance of Deep Learning-Based Super-Resolution for Ultrasound Contrast Imaging

Resolving arterial flows is essential for understanding cardiovascular pathologies, improving diagnosis, and monitoring patient condition. Ultrasound contrast imaging uses microbubbles to enhance the scattering of the blood pool, allowing for real-time visualization of blood flow. Recent developments in vector flow imaging further expand the imaging capabilities of ultrasound by temporally resolving fast arterial flow. The next obstacle to overcome is the lack of spatial resolution. Super-resolved ultrasound images can be obtained by deconvolving radiofrequency (RF) signals before beamforming, breaking the link between resolution and pulse duration. Convolutional neural networks (CNNs) can be trained to locally estimate the deconvolution kernel and consequently super-localize the microbubbles directly within the RF signal. However, microbubble contrast is highly nonlinear, and the potential of CNNs in microbubble localization has not yet been fully exploited. Assessing deep learning-based deconvolution performance for non-trivial imaging pulses is therefore essential for successful translation to a practical setting, where the signal-to-noise ratio is limited, and transmission schemes should comply with safety guidelines. In this study, we train CNNs to deconvolve RF signals and localize the microbubbles driven by harmonic pulses, chirps, or delay-encoded pulse trains. Furthermore, we discuss potential hurdles for in-vitro and in-vivo super-resolution by presenting preliminary experimental results. We find that, whereas the CNNs can accurately localize microbubbles for all pulses, a short imaging pulse offers the best performance in noise-free conditions. However, chirps offer a comparable performance without noise, but are more robust to noise and outperform all other pulses in low-signal-to-noise ratio conditions.

physics.med-ph↗

arXiv2025

Learning Hemodynamic Scalar Fields on Coronary Artery Meshes: A Benchmark of Geometric Deep Learning Models

Coronary artery disease, caused by the narrowing of coronary vessels due to atherosclerosis, is the leading cause of death worldwide. The diagnostic gold standard, fractional flow reserve (FFR), measures the trans-stenotic pressure ratio during maximal vasodilation but is invasive and costly. This has driven the development of virtual FFR (vFFR) using computational fluid dynamics (CFD) to simulate coronary flow. Geometric deep learning algorithms have shown promise for learning features on meshes, including cardiovascular research applications. This study empirically analyzes various backends for predicting vFFR fields in coronary arteries as CFD surrogates, comparing six backends for learning hemodynamics on meshes using CFD solutions as ground truth. The study has two parts: i) Using 1,500 synthetic left coronary artery bifurcations, models were trained to predict pressure-related fields for vFFR reconstruction, comparing different learning variables. ii) Using 427 patient-specific CFD simulations, experiments were repeated focusing on the best-performing learning variable from the synthetic dataset. Most backends performed well on the synthetic dataset, especially when predicting pressure drop over the manifold. Transformer-based backends outperformed others when predicting pressure and vFFR fields and were the only models achieving strong performance on patient-specific data, excelling in both average per-point error and vFFR accuracy in stenotic lesions. These results suggest geometric deep learning backends can effectively replace CFD for simple geometries, while transformer-based networks are superior for complex, heterogeneous datasets. Pressure drop was identified as the optimal network output for learning pressure-related fields.

eess.IV↗