Computational models are continuously integrated in the clinical space, where they support clinicians in disease diagnosis, prognosis, and prevention strategies. While assisting in clinical space, these computational models frequently use deterministic approaches, where the inherent (aleatoric) variability of input parameters is ignored. This questions the credibility and often hinders the clinical adoption of these computational models. Therefore, in this study, we introduced uncertainty quantification in the computational fluid dynamics models of the left main coronary artery to analyze the influence of input hemodynamics parameters on wall shear stress (WSS). UncertainSCI was used, where an emulator was built using polynomial chaos expansion between the input parameters and the output quantity of interest, and the output sensitivities and statistics were directly extracted from the emulator. The uncertainty-informed framework was first applied to an analytical solution of the Navier-Stokes equation (Poiseuille flow) and then to a patient-specific model of the left main coronary artery. Different input hemodynamics parameters are considered, such as pressure, viscosity, density, velocity, and radius, whereas wall shear stress was considered as our output quantity of interest. The results suggest that velocity dominated the variability in WSS in the analytical model (~79%), whereas viscosity dominated in the patient-specific model (~59%). The results further suggest that out of all the Sobol indices interactions, unary interactions were the most dominant ones, contributing ~93.2% and ~99% for the analytical and patient-specific model, respectively. This study will enhance confidence in computational models, facilitating their adoption in the clinical space to improve decision-making for coronary artery disease diagnosis, prognosis, and therapeutic strategies. 

Monitoring HPC systems yields valuable insights into user behavior, aiding resource management, collaborative research, and software design. However, privacy concerns raise the barrier for real-world HPC trace sharing between HPC facilities and researchers. Traditional anonymization methods fall short as user behavior remains identifiable. To address this, we propose a robust toolset for privacy protection of HPC traces using Differential Privacy (DP). Our toolset offers a set of DP algorithms, metrics, and visualizations to empower HPC operators to protect users’ sensitive information under a privacy protection guarantee. We evaluated our toolset over real HPC systems traces for different parameters and data aggregations. Moreover, we show that machine learning models trained on privacy-preserved logs maintain accuracy compared to real data, which supports data publishing and sharing across different computing facilities.

Scientific machine learning has enabled the extraction of physical insights from high-dimensional spatiotemporal flow data using linear and nonlinear dimensionality reduction techniques. Despite these advances, achieving interpretability within the latent space remains a challenge. To address this, we propose the DIfferentiable Autoencoding Neural Operator (DIANO), a deterministic autoencoding neural operator framework that constructs physically interpretable latent spaces for both dimensional and geometric reduction, with the provision to enforce differential governing equations directly within the latent space. Built upon neural operators, DIANO compresses high-dimensional input functions into a low-dimensional latent space via spatial coarsening through an encoding neural operator and subsequently reconstructs the original inputs using a decoding neural operator through spatial refinement. We assess DIANO's latent space interpretability and performance in dimensionality reduction against baseline models, including the Convolutional Neural Operator and standard autoencoders. Furthermore, a fully differentiable partial differential equation (PDE) solver is developed and integrated within the latent space, enabling the temporal advancement of both high- and low-fidelity PDEs, thereby embedding physical priors into the latent dynamics. We further investigate various PDE formulations, including the 2D unsteady advection-diffusion and the 3D Pressure-Poisson equation, to examine their influence on shaping the latent flow representations. Benchmark problems considered include flow past a 2D cylinder, flow through a 2D symmetric stenosed artery, and a 3D patient-specific coronary artery. These case studies demonstrate DIANO's capability to solve PDEs within a latent space that facilitates both dimensional and geometrical reduction while allowing latent interpretability.

 Laser scanning microscopy enables the acquisition of 3D data in biomedical research. A fundamental challenge in visualizing 3D data is that common flat-panel displays, being 2D in nature, cannot faithfully reproduce light fields. Recent years have witnessed the development of various 3D display technologies. These technologies generally fall into two categories, stereography and holography, depending on the number of perspectives they can simultaneously present. We have integrated support for many commercially available 3D-capable displays into FluoRender, a visualization and analysis system for fluorescence microscopy data. This study investigates the opportunities and challenges of applying various 3D display devices in biological research, focusing on their practical use and potential for broad adoption. We found that 3D display devices, including the HoloLens and the Looking Glass, each have their merits and shortcomings. We predict that the convergence of stereographic and holographic technologies will create powerful tools for visualization and analysis in biological applications.

The CS envisions being the global leader in providing technical information, community engagement, and personalized services to computing professionals worldwide. To advance this vision, at the beginning of 2025 the CS reaffirmed its focus on three strategic goals: G1) engaging members, with particular emphasis on students and early career professionals; G2) engaging industry individuals and organizations; and G3) leading in new technical areas.1 These priorities are pursued through programs and activities guided by three operational themes: T1) empowering the volunteer base, T2) enabling nimbleness in execution, and T3) broadening participation. Originally identified by the 2020 CS Planning Committee as the three-year plan, these goals were extended through 2025 to be further reinforced and embedded across the Society.

Rehabilitation can help promote functional restoration following surgical reconstruction of severe bone injuries, yet prescribed regimens are often conservative due to limited understanding of their impact on healing. This study examined rodent physical activity parameters, including wheel running duration, distance, bout frequency, bout duration, velocity, and rest time, assessing their combined impact on bone healing in 2 and 3 mm segmental defects. Artifical intelligence (AI)-based genetic programming generated high-accuracy nonlinear models, revealing a “goldilocks” phenomenon: some rehabilitation stimulates bone formation, but excessive activity is detrimental. Subject-specific finite element modeling showed that local defect compressive strains decreased only in injuries that achieved union and that the correlation between strain and healing evolved over time. These findings highlight the dynamic healing process, necessitating a subject-specific approach. While bone healing is often thought to be primarily driven by achieving optimal strain magnitudes, our results suggest a more complex reality. Bone healing depends not only on optimal strain magnitude but also on balancing activity and rest, which shifts with injury severity and healing progression. Overall, effective rehabilitation must consider injury stabilization, severity, and healing status while ensuring adequate rest to promote optimal bone regeneration.

Valvular heart disease is prevalent and a major contributor to heart failure. Valve leaflet strain is a promising metric for evaluating the mechanics underlying the initiation and progression of valvular pathology. However, robust and generalizable methods for noninvasively quantifying valvular strain from clinically acquired patient images remain limited. In this work, we present a novel feature-tracking framework for quantifying leaflet strain in atrioventricular valves using 3D echocardiographic images of pediatric and adult patients. Our method demonstrated superior accuracy in the assessment of anatomical deformation and strain of heart valves compared to other point-based approaches, as verified against a finite element benchmark. Further, our approach can robustly track inter-phase deformation of valves across highly variable morphologies without parameter tuning. Our analysis revealed that a median and interquartile range of the 1st principal strain greater than 0.5 is associated with leaflet billow (prolapse). Further investigation of the biomechanical signatures of heart valve disease has the potential to enhance prognostic assessment and longitudinal evaluation of valvular disease. 

Particle-based shape modeling (PSM) is a family of approaches that automatically quantifies shape variability across anatomical cohorts by positioning particles (pseudo landmarks) on shape surfaces in a consistent configuration. Recent advances incorporate implicit radial basis function representations as self-supervised signals to better capture the complex geometric properties of anatomical structures. However, these methods still lack self-adaptivity—that is, the ability to automatically adjust particle configurations to local geometric features of each surface, which is essential for accurately representing complex anatomical variability. This paper introduces two mechanisms to increase surface adaptivity while maintaining consistent particle configurations: (1) a novel neighborhood correspondence loss to enable high adaptivity and (2) a geodesic correspondence algorithm that regularizes optimization to enforce geodesic neighborhood consistency. We evaluate the efficacy and scalability of our approach on challenging datasets, providing a detailed analysis of the adaptivity-correspondence trade-off and benchmarking against existing methods on surface representation accuracy and correspondence metrics.

Algebraic multigrid (AMG) is conventionally applied in a black-box fashion, agnostic to the underlying geometry. In this work, we propose that using geometric information -- when available -- to assist with setting up the AMG hierarchy is beneficial, especially for solving linear systems resulting from high-order finite element discretizations. High-order problems draw considerable interest to both the scientific and engineering communities, but lack efficient solvers, at least open-source codes, tailored for unstructured high-order discretizations targeting large-scale, real-world applications. For geometric multigrid, it is known that using p-coarsening before h-coarsening can provide better scalability, but setting up p-coarsening is non-trivial in AMG. We develop a geometrically informed algebraic multigrid (GIAMG) method, as well as an associated high-performance computing program, which is able to set up a grid hierarchy that includes p-coarsening at the top grids with minimal information of the geometry from the user. A major advantage of using p-coarsening with AMG -- beyond the benefits known in the context of geometric multigrid (GMG) -- is the increased sparsification of coarse grid operators. We extensively evaluate GIAMG by testing on the 3D Helmholtz and incompressible flow problems, and demonstrate mesh-independent convergence, and excellent parallel scalability. We also compare the performance of GIAMG with existing AMG packages, including Hypre and ML. 

Machine learning models often rely on simple spurious features – patterns in training data that correlate with targets but are not causally related to them, like image backgrounds in foreground classification. This reliance typically leads to imbalanced test performance across minority and majority groups. In this work, we take a closer look at the fundamental cause of such imbalanced performance through the lens of memorization, which refers to the ability to predict accurately on atypical examples (minority groups) in the training set but failing in achieving the same accuracy in the testing set. This paper systematically shows the ubiquitous existence of spurious features in a small set of neurons within the network, providing the first-ever evidence that memorization may contribute to imbalanced group performance. Through three experimental sources of converging empirical evidence, we find the property of a small subset of neurons or channels in memorizing minority group information. Inspired by these findings, we articulate the hypothesis: the imbalanced group performance is a byproduct of “noisy” spurious memorization confined to a small set of neurons. To further substantiate this hypothesis, we show that eliminating these unnecessary spurious memorization patterns via a novel framework during training can significantly affect the model performance on minority groups. Our experimental results across various architectures and benchmarks offer new insights on how neural networks encode core and spurious knowledge, laying the groundwork for future research in demystifying robustness to spurious correlation.

Machine learning models often rely on simple spurious features – patterns in training data that correlate with targets but are not causally related to them, like image backgrounds in foreground classification. This reliance typically leads to imbalanced test performance across minority and majority groups. In this work, we take a closer look at the fundamental cause of such imbalanced performance through the lens of memorization, which refers to the ability to predict accurately on atypical examples (minority groups) in the training set but failing in achieving the same accuracy in the testing set. This paper systematically shows the ubiquitous existence of spurious features in a small set of neurons within the network, providing the first-ever evidence that memorization may contribute to imbalanced group performance. Through three experimental sources of converging empirical evidence, we find the property of a small subset of neurons or channels in memorizing minority group information. Inspired by these findings, we hypothesize that spurious memorization, concentrated within a small subset of neurons, plays a key role in driving imbalanced group performance. To further substantiate this hypothesis, we show that eliminating these unnecessary spurious memorization patterns via a novel framework during training can significantly affect the model performance on minority groups. Our experimental results across various architectures and benchmarks offer new insights on how neural networks encode core and spurious knowledge, laying the groundwork for future research in demystifying robustness to spurious correlation.

Metastatic spread and therapeutic resistance are the principal causes of cancer mortality. For melanoma, these processes rely on the capacity of cells to switch between transcriptional states. Although targeting transcriptional states pharmacologically is promising, the mechanisms by which melanoma cells switch between states—and how these processes differ from melanocytes—remain poorly understood. Here, we isolate distinct melanoma states with unique phenotypes: a MYC-driven state, essential for tumor initiation yet sensitive to BRAF inhibition, and a dedifferentiated, invasive BRN2-high state enriched in therapy-resistant cells but not directly tumorigenic. Transitions between phenotypes occur through intermediate, more differentiated states. Unexpectedly, the BRN2-high state is also present in melanocytes, whereas the MYC state is exclusive to melanoma. Melanoma cells also exhibit an increased frequency of transitions across states. These findings highlight that accelerated phenotypic switching, rather than mere state diversity, is a defining feature of melanoma progression.

Anatomical shape analysis plays a pivotal role in clinical research and hypothesis testing, where the relationship between form and function is paramount. Correspondence-based statistical shape modeling (SSM) facilitates population-level morphometrics but requires a cumbersome, potentially bias-inducing construction pipeline. Recent advancements in deep learning have streamlined this process in inference by providing SSM prediction directly from unsegmented medical images. However, the proposed approaches are fully supervised and require utilizing a traditional SSM construction pipeline to create training data, thus inheriting the associated burdens and limitations. To address these challenges, we introduce a weakly supervised deep learning approach to predict SSM from images using point cloud supervision. Specifically, we propose reducing the supervision associated with the state-of-the-art fully Bayesian variational information bottleneck DeepSSM (BVIB-DeepSSM) model. BVIB-DeepSSM is an effective, principled framework for predicting probabilistic anatomical shapes from images with quantification of both aleatoric and epistemic uncertainties. Whereas the original BVIB-DeepSSM method requires strong supervision in the form of ground truth correspondence points, the proposed approach utilizes weak supervision via point cloud surface representations, which are more readily obtainable. Furthermore, the proposed approach learns correspondence in a completely data-driven manner without prior assumptions about the expected variability in shape cohort. Our experiments demonstrate that this approach yields similar accuracy and uncertainty estimation to the fully supervised scenario while substantially enhancing the feasibility of model training for SSM construction.

Correspondence-based statistical shape modeling (SSM) stands as a powerful technology for morphometric analysis in clinical research. SSM facilitates population-level characterization and quantification of anatomical shapes such as bones and organs, aiding in pathology and disease diagnostics and treatment planning. Despite its potential, SSM remains under-utilized in medical research due to the significant overhead associated with automatic construction methods, which demand complete, aligned shape surface representations. Additionally, optimization-based techniques rely on bias-inducing assumptions or templates and have prolonged inference times as the entire cohort is simultaneously optimized. To overcome these challenges, we introduce Point2SSM++, a principled, self-supervised deep learning approach that directly learns correspondence points from point cloud representations of anatomical shapes. Point2SSM++ is robust to misaligned and inconsistent input, providing SSM that accurately samples individual shape surfaces while effectively capturing population-level statistics. Additionally, we present principled extensions of Point2SSM++ to adapt it for dynamic spatiotemporal and multi-anatomy use cases, demonstrating the broad versatility of the Point2SSM++ framework. Furthermore, we present extensions of Point2SSM++ tailored for dynamic spatiotemporal and multi-anatomy scenarios, showcasing the broad versatility of the framework. Through extensive validation across diverse anatomies, evaluation metrics, and clinically relevant downstream tasks, we demonstrate Point2SSM++’s superiority over existing state-of-the-art deep learning models and traditional approaches. Point2SSM++ substantially enhances the feasibility of SSM generation and significantly broadens its array of potential clinical applications.

This work investigates the use of configuration state imaging together with deep neural networks to develop quantitative MRI techniques for deployment in an interventional setting. A physics modeling technique for inhomogeneous fields and heterogeneous tissues is presented and used to evaluate the theoretical capability of neural networks to estimate parameter maps from configuration state signal data. All tested normalization strategies achieved similar performance in estimating T₂ and T₂*. Varying network architecture and data normalization had substantial impacts on estimated flip angle and T₁, highlighting their importance in developing neural networks to solve these inverse problems. The developed signal modeling technique provides an environment that will enable the development and evaluation of physics-informed machine learning techniques for MR parameter mapping and facilitate the development of quantitative MRI techniques to inform clinical decisions during MR-guided treatments.

Color maps are a commonly used visualization technique in which data are mapped to optical properties, e.g., color or opacity. Color maps, however, do not explicitly convey structures (e.g., positions and scale of features) within data. Topology-based visualizations reveal and explicitly communicate structures underlying data. Although we have a good understanding of what types of features are captured by topological visualizations, our understanding of people’s perception of those features is not. This paper evaluates the sensitivity of topology-based isocontour, Reeb graph, and persistence diagram visualizations compared to a reference color map visualization for synthetically generated scalar fields on 2-manifold triangular meshes embedded in 3D. In particular, we built and ran a human-subject study that evaluated the perception of data features characterized by Gaussian signals and measured how effectively each visualization technique portrays variations of data features arising from the position and amplitude variation of a mixture of Gaussians. For positional feature variations, the results showed that only the Reeb graph visualization had high sensitivity. For amplitude feature variations, persistence diagrams and color maps demonstrated the highest sensitivity, whereas isocontours showed only weak sensitivity. These results take an important step toward understanding which topology-based tools are best for various data and task scenarios and their effectiveness in conveying topological variations as compared to conventional color mapping.

Patient-specific 3D vertebra models are essential for accurately assessing the spinal deformities quantitatively in 3D and for surgical planning, including determining the optimal implant size and 3D positioning. Calibrated biplanar X-rays serve as an alternative to CT scans to generate the 3D models in a weight-bearing standing position. This paper presents an intensity-based 3D/2D registration method for vertebra statistical shape model (VSSM), incorporating two key elements: the particle-based shape modeling and an image domain transfer for efficient image matching. In the 3D/3D setting, the VSSMs reach a surface reconstruction error of less than 0.5 mm. For 3D reconstruction from biplanar X-rays, the root mean square point-to-surface are 1.05 mm for L1 to L4 vertebrae and 1.6 mm for the L5 vertebra. The particle-based VSSMs offer a significant balance between the model compactness and the reconstruction error, which is advantageous for deformable 3D/2D registration.

Transformers have emerged as the state-of-the-art architecture in medical image registration, outperforming convolutional neural networks (CNNs) by addressing their limited receptive fields and overcoming gradient instability in deeper models. Despite their success, transformer-based models require substantial resources for training, including data, memory, and computational power, which may restrict their applicability for end users with limited resources. In particular, existing transformer-based 3D image registration architectures face three critical gaps that challenge their efficiency and effectiveness. Firstly, while mitigating the quadratic complexity of full attention by focusing on local regions, window-based attention mechanisms often fail to adequately integrate local and global information. Secondly, feature similarities across attention heads that were recently found in multi-head attention architectures indicate a significant computational redundancy, suggesting that the capacity of the network could be better utilized to enhance performance. Lastly, the granularity of tokenization, a key factor in registration accuracy, presents a trade-off; smaller tokens improve detail capture at the cost of higher computational complexity, increased memory demands, and a risk of overfitting. Here, we propose EfficientMorph, a transformer-based architecture for unsupervised 3D image registration. It optimizes the balance between local and global attention through a plane-based attention mechanism, reduces computational redundancy via cascaded group attention, and captures fine details without compromising computational efficiency, thanks to a Hi-Res tokenization strategy complemented by merging operations. We compare the effectiveness of EfficientMorph on two public datasets, OASIS and IXI, against other state-of-the-art models. Notably, EfficientMorph sets a new benchmark for performance on the OASIS dataset with ∼16-27× fewer parameters.

Learning long-range interactions (LRI) between distant nodes is crucial for many graph learning tasks. Predominant graph neural networks (GNNs) rely on local message passing and struggle to learn LRI. In this paper, we propose DRGNN to learn LRI leveraging monotone operator theory. DRGNN contains two key components: (1) we use a full node similarity matrix beyond adjacency matrix – drawing inspiration from the personalized PageRank matrix – as the aggregation matrix for message passing, and (2) we implement message-passing on graphs using Douglas-Rachford splitting to circumvent prohibitive matrix inversion. We demonstrate that DRGNN surpasses various advanced GNNs, including Transformer-based models, on several benchmark LRI learning tasks arising from different application domains, highlighting its efficacy in learning LRI. Code is available at \urlhttps://github.com/Utah-Math-Data-Science/PR-inspired-aggregation.

This paper proposes a novel deep symbolic regression approach to enhance the robustness and interpretability of data-driven mathematical expression discovery. Despite the success of the state-of-the-art method, DSR, it is built on recurrent neural networks, purely guided by data fitness, and potentially meet tail barriers, which can zero out the policy gradient and cause inefficient model updates. To overcome these limitations, we use transformers in conjunction with breadth-first-search to improve the learning performance. We use Bayesian information criterion (BIC) as the reward function to explicitly account for the expression complexity and optimize the trade-off between interpretability and data fitness. We propose a modified risk-seeking policy that not only ensures the unbiasness of the gradient, but also removes the tail barriers, thus ensuring effective updates from top performers. Through a series of benchmarks and systematic experiments, we demonstrate the advantages of our approach.