arXiv subjects

Jürgen Hesser

Publications and source records attributed to Jürgen Hesser.

At least 19 recordsLinked to original sources

arXiv2026

Constrained estimation of rotational invariants of the cumulant expansion (RICE) for rapid tensor-valued diffusion MRI

Purpose: To complement 1.5-minute measurements of common tensor-valued diffusion MRI (dMRI) markers with rapid constrained fitting. Methods: Fast dMRI protocols for obtaining rotational invariants of the cumulant expansion (RICE) were paired with constrained weighted linear least squares (CWLLS) to stabilize the more fragile WLLS fit. A compact constraint set was formulated, including a novel mean-dependent upper bound on total diffusional variance. Evaluation used diffusion tensor distribution (DTD) simulations, healthy-volunteer data with a resolution-dependent SNR experiment, and a glioma patient dataset. A 5-minute q-space trajectory imaging (QTI) protocol served as a reference. Results: Across experiments, CWLLS reduced unphysical estimates and fit outliers in parameters such as microscopic FA and isotropic diffusivity variance. In simulations, it narrowed error distributions most clearly in the CSF-dominant case, while some metrics showed a bias-variance trade-off. In vivo, CWLLS removed negative variance estimates, truncated out-of-bounds tails, and reduced artifacts in fluid-contaminated voxels while preserving anatomical contrast. It also retained more stable maps than WLLS at higher resolution, although both estimators degraded in the lowest-SNR setting. Notably, the new mean-dependent variance bound was violated in 15.4% of voxels in the patient dataset, accounting for nearly half of the 32.7% that violated at least one constraint. Healthy-volunteer benchmarking showed that CWLLS completed in under 30 seconds. The constrained QTI fit required 72 minutes, making CWLLS 160 times faster. Conclusion: CWLLS for fast RICE yielded high-quality parameter maps at an online-ready computational cost. This may enhance the reliability of dMRI tissue characterization and strengthen the path toward clinical translation.

physics.med-ph↗

arXiv2026

Frequency-Structured Field Learning for Light-Field Disparity Estimation

Light-field disparity estimation requires global consistency in smooth or textureless regions and local precision near occlusion boundaries, thin structures, and abrupt depth transitions. Existing methods address these requirements through EPI matching, cost-volume or focal-stack construction, view aggregation, or direct convolutional regression, often relying on local windows, discrete disparity hypotheses, memory-intensive volumes, or attention-based aggregation. We instead formulate disparity estimation at the field level, predicting disparity from globally and locally updated EPI-derived latent features without explicitly constructing a disparity volume. We introduce FreqLF, an EPI-guided Fourier-local framework that encodes angular parallax cues from horizontal and vertical EPI stacks together with central-view appearance features. These cues are projected into a latent field and updated through stacked hybrid Fourier-local layers. Fourier low-mode updates enable global feature interaction, while local convolutions preserve spatial variations needed for fine disparity detail. A coordinate-conditioned Gaussian-mixture decoder then predicts disparity, using the mixture mean as the final estimate. Experiments on the HCI 4D Light Field Benchmark show that FreqLF approaches the accuracy of strong supervised baselines while avoiding explicit cost-volume construction in the base model. Ablations confirm the complementary roles of the Fourier and local branches, and scaling experiments demonstrate practical behavior across spatial resolutions. These results suggest that Fourier-local latent field learning is a competitive alternative for light-field disparity estimation. The code will be published soon.

cs.CV↗

arXiv2026

DINE: Distance Is Not Enough -- Learning Global Deformation Priors for Robust Soft-Tissue Point Cloud Registration

Non-rigid point cloud registration is central to soft-tissue shape analysis, but large deformations, noise, and outliers make correspondence estimation challenging. Most learning-based methods rely on local objectives such as Chamfer distance, which encourage point-wise proximity but do not constrain the global plausibility of the predicted deformation field. We address this limitation with DINE, a maximum a posteriori framework that augments distance-based registration with a learned statistical prior over displacement vector fields. DINE is applied to two registration backbones, Robust-DefReg and DefTransNet, using a two-stage strategy: a first-stage model is trained with Chamfer distance, its predicted deformation fields are used to estimate a prior, and the model is then refined with a combined distance and negative log-prior objective. We compare a full-field PCA Gaussian prior with a per-vector normalizing-flow prior. Experiments on DeformedTissue and SynBench show lower mean Chamfer distance under deformation and corruption. On DeformedTissue, DINE-PCA reduces Chamfer distance by approximately 27--69\% relative to the corresponding Stage-1 backbone across deformation levels, and improves robustness by up to 66\% for outliers and 83\% for Gaussian noise. On SynBench, improvements are modest at the smallest deformation levels and reach approximately 59--79\% from moderate to severe deformation. These results suggest that global deformation plausibility is an important constraint for reliable soft-tissue point cloud registration. (The code will be published soon.)

cs.CV↗

arXiv2026

Local Conformal Predictions for Calibrated Surrogates

Neural network surrogates for LHC scattering amplitudes require trustworthy uncertainty estimates, a challenging task given the non-Gaussian systematics. We target it using conformal prediction, a distribution-free post-processing to complement trained surrogates with calibrated uncertainties. We find that standard conformal predictions struggle to provide locally calibrated uncertainties. This leads us to introduce FALCON, a novel conformal prediction method that learns locally calibrated confidence intervals. Our simple examples illustrate the power of distribution-free uncertainty quantification for ultra-fast event generation at the LHC.

hep-ph↗

arXiv2026

Adaptive aggregation of Monte Carlo augmented decomposed filters for efficient group-equivariant convolutional neural network

Group-equivariant convolutional neural networks (G-CNN) heavily rely on parameter sharing to increase CNN's data efficiency and performance. However, the parameter-sharing strategy greatly increases the computational burden for each added parameter, which hampers its application to deep neural network models. In this paper, we address these problems by proposing a non-parameter-sharing approach for group equivariant neural networks. The proposed methods adaptively aggregate a diverse range of filters by a weighted sum of stochastically augmented decomposed filters. We give theoretical proof about how the group equivariance can be achieved by our methods. Our method applies to both continuous and discrete groups, where the augmentation is implemented using Monte Carlo sampling and bootstrap resampling, respectively. Our methods also serve as an efficient extension of standard CNN. The experiments show that our method outperforms parameter-sharing group equivariant networks and enhances the performance of standard CNNs in image classification and denoising tasks, by using suitable filter bases to build efficient lightweight networks. The code is available at https://github.com/ZhaoWenzhao/MCG_CNN.

cs.CV↗

arXiv2026

Exploring Drug Safety Through Knowledge Graphs: Protein Kinase Inhibitors as a Case Study

Adverse Drug Reactions (ADRs) are a leading cause of morbidity and mortality. Existing prediction methods rely mainly on chemical similarity, machine learning on structured databases, or isolated target profiles, but often fail to integrate heterogeneous, partly unstructured evidence effectively. We present a knowledge graph-based framework that unifies diverse sources, drug-target data (ChEMBL), clinical trial literature (PubMed), trial metadata (ClinicalTrials.gov), and post-marketing safety reports (FAERS) into a single evidence-weighted bipartite network of drugs and medical conditions. Applied to 400 protein kinase inhibitors, the resulting network enables contextual comparison of efficacy (HR, PFS, OS), phenotypic and target similarity, and ADR prediction via target-to-adverse-event correlations. A non-small cell lung cancer case study correctly highlights established and candidate drugs, target communities (ERbB, ALK, VEGF), and tolerability differences. Designed as an orthogonal, extensible analysis and search tool rather than a replacement for current models, the framework excels at revealing complex patterns, supporting hypothesis generation, and enhancing pharmacovigilance. Code and data are publicly available at https://github.com/davidjackson99/PKI_KG.

q-bio.BM↗

arXiv2026

Efficient 3D affinely equivariant CNNs with adaptive fusion of augmented spherical Fourier-Bessel bases

Filter-decomposition-based group equivariant convolutional neural networks (CNNs) have shown promising stability and data efficiency for 3D image feature extraction. However, these networks, which rely on parameter sharing and discrete transformation groups, often underperform in modern deep neural network architectures for processing volumetric images with dense 3D textures, such as the common 3D medical images. To address these limitations, this paper presents an efficient non-parameter-sharing continuous 3D affine group equivariant neural network for volumetric images. This network uses an adaptive aggregation of Monte Carlo augmented spherical Fourier-Bessel filter bases to improve the efficiency and flexibility of 3D group equivariant CNNs for volumetric data. Unlike existing methods that focus only on angular orthogonality in filter bases, the introduced spherical Bessel Fourier filter base incorporates both angular and radial orthogonality to improve feature extraction. Experiments on four medical image segmentation datasets and two seismic datasets show that the proposed methods achieve better affine group equivariance and superior segmentation accuracy than existing 3D group equivariant convolutional neural network layers, significantly improving the training stability and data efficiency of conventional CNN layers (at 0.05 significance level). The code is available at https://github.com/ZhaoWenzhao/WMCSFB.

cs.CV↗

arXiv2025

A Systematic Review of Conformal Inference Procedures for Treatment Effect Estimation: Methods and Challenges

Treatment effect estimation is essential for informed decision-making in many fields such as healthcare, economics, and public policy. While flexible machine learning models have been widely applied for estimating heterogeneous treatment effects, quantifying the inherent uncertainty of their point predictions remains an issue. Recent advancements in conformal prediction address this limitation by allowing for inexpensive computation, as well as distribution shifts, while still providing frequentist, finite-sample coverage guarantees under minimal assumptions for any point-predictor model. This advancement holds significant potential for improving decision-making in especially high-stakes environments. In this work, we perform a systematic review regarding conformal prediction methods for treatment effect estimation and provide for both the necessary theoretical background. Through a systematic filtering process, we select and analyze eleven key papers, identifying and describing current state-of-the-art methods in this area. Based on our findings, we propose directions for future research.

cs.LG↗

arXiv2025

A baseline for machine-learning-based hepatocellular carcinoma diagnosis using multi-modal clinical data

The objective of this paper is to provide a baseline for performing multi-modal data classification on a novel open multimodal dataset of hepatocellular carcinoma (HCC), which includes both image data (contrast-enhanced CT and MRI images) and tabular data (the clinical laboratory test data as well as case report forms). TNM staging is the classification task. Features from the vectorized preprocessed tabular data and radiomics features from contrast-enhanced CT and MRI images are collected. Feature selection is performed based on mutual information. An XGBoost classifier predicts the TNM staging and it shows a prediction accuracy of $0.89 \pm 0.05$ and an AUC of $0.93 \pm 0.03$. The classifier shows that this high level of prediction accuracy can only be obtained by combining image and clinical laboratory data and therefore is a good example case where multi-model classification is mandatory to achieve accurate results.

cs.CV↗

arXiv2024

3D-Localization of Single Point-Like Gamma Sources with a Coded Aperture Camera

3D-localization of gamma sources has the potential to improve the outcome of radio-guided surgery. The goal of this paper is to analyze the localization accuracy for point-like sources with a single coded aperture camera. We both simulated and measured a point-like $^{241}$Am source at $17$ positions distributed within the field of view of an experimental gamma camera. The setup includes a 0.11mm thick tungsten sheet with a MURA mask of rank $31$ and pinholes of $0.08$mm in diameter and a detector based on the photon counting readout circuit Timepix3. Two methods, namely an iterative search (ISL) including either a symmetric Gaussian fitting or an exponentially modified Gaussian fitting (EMG) and a center of mass method were compared to estimate the 3D source position. Considering the decreasing axial resolution with source-to-mask distance, the EMG improved the results by a factor of $4$ compared to the Gaussian fitting based on the simulated data. Overall, we obtained a mean localization error of $0.77$mm on the simulated and $2.64$mm on the experimental data in the imaging range of $20$mm to $100$ mm. This paper shows that despite the low axial resolution, point-like sources in the nearfield can be localized as well as with more sophisticated imaging devices such as stereo cameras. The influence of the source size and the photon count on the imaging and localization accuracy remains an important issue for further research. The acquired datasets and the localization methods of this research are publicly available on GitHub at "https://zenodo.org/records/11449544".

physics.med-ph↗

arXiv2024

Assessment of the Axial Resolution of a Compact Gamma Camera with Coded Aperture Collimator

Purpose: Handheld gamma cameras with coded aperture collimators are under investigation for intraoperative imaging in nuclear medicine. Coded apertures are a promising collimation technique for applications such as lymph node localization due to their high sensitivity and the possibility of 3D imaging. We evaluated the axial resolution and computational performance of two reconstruction methods. Methods: An experimental gamma camera was set up consisting of the pixelated semiconductor detector Timepix3 and MURA mask of rank $31$ with round holes of $0.08$mm in diameter in a $0.11$mm thick Tungsten sheet. A set of measurements was taken where a point-like gamma source was placed centrally at $21$ different positions within the range of $12$ to $100$mm. For each source position, the detector image was reconstructed in $0.5$mm steps around the true source position, resulting in an image stack. The axial resolution was assessed by the full width at half maximum (FWHM) of the contrast-to-noise ratio (CNR) profile along the z-axis of the stack. Two reconstruction methods were compared: MURA Decoding and a 3D maximum likelihood expectation maximization algorithm (3D-MLEM). Results: While taking $4{,}400$ times longer in computation, 3D-MLEM yielded a smaller axial FWHM and a higher CNR. The axial resolution degraded from $5.3$mm and $1.8$mm at $12$mm to $42.2$mm and $13.5$mm at $100$mm for MURA Decoding and 3D-MLEM respectively. Conclusion: Our results show that the coded aperture enables the depth estimation of single point-like sources in the near field. Here, 3D-MLEM offered a better axial resolution but was computationally much slower than MURA Decoding, whose reconstruction time is compatible with real-time imaging.

physics.med-ph↗

arXiv2023

Ensemble Kalman Inversion for Image Guided Guide Wire Navigation in Vascular Systems

This paper addresses the challenging task of guide wire navigation in cardiovascular interventions, focusing on the parameter estimation of a guide wire system using Ensemble Kalman Inversion (EKI) with a subsampling technique. The EKI uses an ensemble of particles to estimate the unknown quantities. However since the data misfit has to be computed for each particle in each iteration, the EKI may become computationally infeasible in the case of high-dimensional data, e.g. high-resolution images. This issue can been addressed by randomised algorithms that utilize only a random subset of the data in each iteration. We introduce and analyse a subsampling technique for the EKI, which is based on a continuous-time representation of stochastic gradient methods and apply it to on the parameter estimation of our guide wire system. Numerical experiments with real data from a simplified test setting demonstrate the potential of the method.

math.NA↗

arXiv2023

Simulation Study on Super-Resolution for Coded Aperture Gamma Imaging

Coded Aperture Imaging (CAI) has been proposed as an alternative collimation technique in nuclear imaging. To maximize spatial resolution small pinholes in the coded aperture mask are required. However, a high-resolution detector is needed to correctly sample the point spread function (PSF) to keep the Nyquist-Shannon sampling theorem satisfied. The disadvantage of smaller pixels, though, is the resulting higher Poisson noise. Thus, the aim of this paper was to investigate if sufficiently accurate CAI reconstruction is achievable with a detector which undersamples the PSF. With the Monte Carlo simulation framework TOPAS a test image with multiple spheres of different diameter was simulated based on the setup of an experimental gamma camera from previous work. Additionally, measured phantom data were acquired. The captured detector images were converted to low-resolution images of different pixel sizes according to the super-resolution factor $k$. Multiple analytical reconstruction methods and a Machine Learning approach were compared based on the contrast-to-noise ratio (CNR). We show, that all reconstruction methods are able to reconstruct both the test image and the measured phantom data for $k \leq 7$. With a synthetic high-resolution PSF and upsampling the simulated low-resolution detector image by bilinear interpolation the CNR can be kept approximately constant. Results of this simulation study and additional validation on measured phantom data indicate that an undersampling detector can be combined with small aperture holes. However, further experiments need to be conducted.

physics.med-ph↗

arXiv2023

Robust-DefReg: A Robust Deformable Point Cloud Registration Method based on Graph Convolutional Neural Networks

Point cloud registration is a fundamental problem in computer vision that aims to estimate the transformation between corresponding sets of points. Non-rigid registration, in particular, involves addressing challenges including various levels of deformation, noise, outliers, and data incompleteness. This paper introduces Robust-DefReg, a robust non-rigid point cloud registration method based on graph convolutional networks (GCNNs). Robust-DefReg is a coarse-to-fine registration approach within an end-to-end pipeline, leveraging the advantages of both coarse and fine methods. The method learns global features to find correspondences between source and target point clouds, to enable appropriate initial alignment, and subsequently fine registration. The simultaneous achievement of high accuracy and robustness across all challenges is reported less frequently in existing studies, making it a key objective of the Robust-DefReg method. The proposed method achieves high accuracy in large deformations while maintaining computational efficiency. This method possesses three primary attributes: high accuracy, robustness to different challenges, and computational efficiency. The experimental results show that the proposed Robust-DefReg holds significant potential as a foundational architecture for future investigations in non-rigid point cloud registration. The source code of Robust-DefReg is available.

cs.CV↗

arXiv2022

The Liver Tumor Segmentation Benchmark (LiTS)

In this work, we report the set-up and results of the Liver Tumor Segmentation Benchmark (LiTS), which was organized in conjunction with the IEEE International Symposium on Biomedical Imaging (ISBI) 2017 and the International Conferences on Medical Image Computing and Computer-Assisted Intervention (MICCAI) 2017 and 2018. The image dataset is diverse and contains primary and secondary tumors with varied sizes and appearances with various lesion-to-background levels (hyper-/hypo-dense), created in collaboration with seven hospitals and research institutions. Seventy-five submitted liver and liver tumor segmentation algorithms were trained on a set of 131 computed tomography (CT) volumes and were tested on 70 unseen test images acquired from different patients. We found that not a single algorithm performed best for both liver and liver tumors in the three events. The best liver segmentation algorithm achieved a Dice score of 0.963, whereas, for tumor segmentation, the best algorithms achieved Dices scores of 0.674 (ISBI 2017), 0.702 (MICCAI 2017), and 0.739 (MICCAI 2018). Retrospectively, we performed additional analysis on liver tumor detection and revealed that not all top-performing segmentation algorithms worked well for tumor detection. The best liver tumor detection method achieved a lesion-wise recall of 0.458 (ISBI 2017), 0.515 (MICCAI 2017), and 0.554 (MICCAI 2018), indicating the need for further research. LiTS remains an active benchmark and resource for research, e.g., contributing the liver-related segmentation tasks in \url{http://medicaldecathlon.com/}. In addition, both data and online evaluation are accessible via \url{www.lits-challenge.com}.

cs.CV↗

arXiv2020

Mask Mining for Improved Liver Lesion Segmentation

We propose a novel procedure to improve liver and lesion segmentation from CT scans for U-Net based models. Our method extends standard segmentation pipelines to focus on higher target recall or reduction of noisy false-positive predictions, boosting overall segmentation performance. To achieve this, we include segmentation errors into a new learning process appended to the main training setup, allowing the model to find features which explain away previous errors. We evaluate this on semantically distinct architectures: cascaded two- and three-dimensional as well as combined learning setups for multitask segmentation. Liver and lesion segmentation data are provided by the Liver Tumor Segmentation challenge (LiTS), with an increase in dice score of up to 2 points.

eess.IV↗

arXiv2019

Liver Lesion Segmentation with slice-wise 2D Tiramisu and Tversky loss function

At present, lesion segmentation is still performed manually (or semi-automatically) by medical experts. To facilitate this process, we contribute a fully-automatic lesion segmentation pipeline. This work proposes a method as a part of the LiTS (Liver Tumor Segmentation Challenge) competition for ISBI 17 and MICCAI 17 comparing methods for automatics egmentation of liver lesions in CT scans. By utilizing cascaded, densely connected 2D U-Nets and a Tversky-coefficient based loss function, our framework achieves very good shape extractions with high detection sensitivity, with competitive scores at time of publication. In addition, adjusting hyperparameters in our Tversky-loss allows to tune the network towards higher sensitivity or robustness.

cs.CV↗

arXiv2019

Automated Multiscale 3D Feature Learning for Vessels Segmentation in Thorax CT Images

We address the vessel segmentation problem by building upon the multiscale feature learning method of Kiros et al., which achieves the current top score in the VESSEL12 MICCAI challenge. Following their idea of feature learning instead of hand-crafted filters, we have extended the method to learn 3D features. The features are learned in an unsupervised manner in a multi-scale scheme using dictionary learning via least angle regression. The 3D feature kernels are further convolved with the input volumes in order to create feature maps. Those maps are used to train a supervised classifier with the annotated voxels. In order to process the 3D data with a large number of filters a parallel implementation has been developed. The algorithm has been applied on the example scans and annotations provided by the VESSEL12 challenge. We have compared our setup with Kiros et al. by running their implementation. Our current results show an improvement in accuracy over the slice wise method from 96.66$\pm$1.10% to 97.24$\pm$0.90%.

cs.CV↗