SCIENTIFIC COMPUTING AND IMAGING INSTITUTE
at the University of Utah

Practical tensor data is often along with time information. Most existing temporal decomposition approaches estimate a set of fixed factors for the objects in each tensor mode, and hence cannot capture the temporal evolution of the objects' representation. More important, we lack an effective approach to capture such evolution from streaming data, which is common in real-world applications. To address these issues, we propose Streaming Factor Trajectory Learning for temporal tensor decomposition. We use Gaussian processes (GPs) to model the trajectory of factors so as to flexibly estimate their temporal evolution. To address the computational challenges in handling streaming data, we convert the GPs into a state-space prior by constructing an equivalent stochastic differential equation (SDE). We develop an efficient online filtering algorithm to estimate a decoupled running posterior of the involved factor states upon receiving new data. The decoupled estimation enables us to conduct standard Rauch-Tung-Striebel smoothing to compute the full posterior of all the trajectories in parallel, without the need for revisiting any previous data. We have shown the advantage of SFTL in both synthetic tasks and real-world applications.

Machine learning based solvers have garnered much attention in physical simulation and scientific computing, with a prominent example, physics-informed neural networks (PINNs). However, PINNs often struggle to solve high-frequency and multi-scale PDEs, which can be due to the spectral bias during neural network training. To address this problem, we resort to the Gaussian process (GP) framework. To flexibly capture the dominant frequencies, we model the power spectrum of the PDE solution with a student t mixture or Gaussian mixture. We then apply inverse Fourier transform to obtain the covariance function (according to the Wiener-Khinchin theorem). The covariance derived from the Gaussian mixture spectrum corresponds to the known spectral mixture kernel. We are the first to discover its rationale and effectiveness for PDE solving. Next, we estimate the mixture weights in the log domain, which we show is equivalent to placing a Jeffreys prior. It automatically induces sparsity, prunes excessive frequencies, and adjusts the remaining toward the ground truth. Third, to enable efficient and scalable computation on massive collocation points, which are critical to capture high frequencies, we place the collocation points on a grid, and multiply our covariance function at each input dimension. We use the GP conditional mean to predict the solution and its derivatives so as to fit the boundary condition and the equation itself. As a result, we can derive a Kronecker product structure in the covariance matrix. We use Kronecker product properties and multilinear algebra to greatly promote computational efficiency and scalability, without any low-rank approximations. We show the advantage of our method in systematic experiments.

This paper presents an overview of an NSF Research Experience for Undergraduate (REU) Site on Trust and Reproducibility of Intelligent Computation, delivered by faculty and graduate students in the Kahlert School of Computing at University of Utah. The chosen themes bring together several concerns for the future in producing computational results that can be trusted: secure, reproducible, based on sound algorithmic foundations, and developed in the context of ethical considerations. The research areas represented by student projects include machine learning, high-performance computing, algorithms and applications, computer security, data science, and human-centered computing. In the first four weeks of the program, the entire student cohort spent their mornings in lessons from experts in these crosscutting topics, and used one-of-a-kind research platforms operated by the University of Utah, namely NSF-funded CloudLab and POWDER facilities; reading assignments, quizzes, and hands-on exercises reinforced the lessons. In the subsequent five weeks, lectures were less frequent, as students branched into small groups to develop their research projects. The final week focused on a poster presentation and final report. Through describing our experiences, this program can serve as a model for preparing a future workforce to integrate machine learning into trustworthy and reproducible applications.

Forward simulation-based uncertainty quantification that studies the output distribution of quantities of interest (QoI) is a crucial component for computationally robust statistics and engineering. There is a large body of literature devoted to accurately assessing statistics of QoI, and in particular, multilevel or multifidelity approaches are known to be effective, leveraging cost-accuracy tradeoffs between a given ensemble of models. However, effective algorithms that can estimate the full distribution of outputs are still under active development. In this paper, we introduce a general multifidelity framework for estimating the cumulative distribution functions (CDFs) of vector-valued QoI associated with a high-fidelity model under a budget constraint. Given a family of appropriate control variates obtained from lower fidelity surrogates, our framework involves identifying the most cost-effective model subset and then using it to build an approximate control variates estimator for the target CDF. We instantiate the framework by constructing a family of control variates using intermediate linear approximators and rigorously analyze the corresponding algorithm. Our analysis reveals that the resulting CDF estimator is uniformly consistent and budget-asymptotically optimal, with only mild moment and regularity assumptions. The approach provides a robust multifidelity CDF estimator that is adaptive to the available budget, does not require a priori knowledge of cross-model statistics or model hierarchy, and is applicable to general output dimensions. We demonstrate the efficiency and robustness of the approach using several test examples.

Metabolic networks are interconnected and influence diverse cellular processes. The protein-metabolite interactions that mediate these networks are frequently low affinity and challenging to systematically discover. We developed mass spectrometry integrated with equilibrium dialysis for the discovery of allostery systematically (MIDAS) to identify such interactions. Analysis of 33 enzymes from human carbohydrate metabolism identified 830 protein-metabolite interactions, including known regulators, substrates, and products as well as previously unreported interactions. We functionally validated a subset of interactions, including the isoform-specific inhibition of lactate dehydrogenase by long-chain acyl–coenzyme A. Cell treatment with fatty acids caused a loss of pyruvate-lactate interconversion dependent on lactate dehydrogenase isoform expression. These protein-metabolite interactions may contribute to the dynamic, tissue-specific metabolic flexibility that enables growth and survival in an ever-changing nutrient environment. Understanding how metabolic state influences cellular processes requires systematic analysis of low-affinity interactions of metabolites with proteins. Hicks et al. describe a method called MIDAS (mass spectrometry integrated with equilibrium dialysis for the discovery of allostery systematically), which allowed them to probe such interactions for 33 enzymes of human carbohydrate metabolism and more than 400 metabolites. The authors detected many known and many new interactions, including regulation of lactate dehydrogenase by ATP and long-chain acyl coenzyme A, which may help to explain known physiological relations between fat and carbohydrate metabolism in different tissues. —LBR A mass spectrometry and dialysis method detects metabolite-protein interactions that help to control physiology.

Scientific simulations and observations using particles have been creating large datasets that require effective and efficient data reduction to store, transfer, and analyze. However, current approaches either compress only small data well while being inefficient for large data, or handle large data but with insufficient compression. Toward effective and scalable compression/decompression of particle positions, we introduce new kinds of particle hierarchies and corresponding traversal orders that quickly reduce reconstruction error while being fast and low in memory footprint. Our solution to compression of large-scale particle data is a flexible block-based hierarchy that supports progressive, random-access, and error-driven decoding, where error estimation heuristics can be supplied by the user. For low-level node encoding, we introduce new schemes that effectively compress both uniform and densely structured particle distributions.

The US Department of Energy (DOE) oversees a system of 17 national laboratories responsible for developing unique scientific capabilities beyond the scope of academic and industrial institutions. These labs strive to keep America at the forefront of discovery and are home to some of the Nation’s best minds and the world’s best scientific and research facilities. Collaborations between national laboratories and academic institutions are critical to develop and recruit talent for the DOE workforce. Academia’s cooperative education model poses challenges for DOE recruitment pipelines centered around traditional internships. This paper discusses a promising DOE recruitment pipeline, the National Nuclear Security Administration’s (NNSA) Predictive Science Academic Alliance Program (PSAAP) initiative. As a part of this, experiences capturing the successes and challenges faced by the University of Utah’s Carbon Capture Multidisciplinary Simulation Center (CCMSC) through their participation in the PSAAP-II initiative are shared. These experiences demonstrate the success of Utah’s PSAAP center as a recruitment pipeline with approximately 43% of CCMSC students going to a national laboratory after graduation. Potential opportunities to strengthen the DOE’s recruitment pipeline are also discussed.

In this study, a systematic review and meta-analysis were conducted to identify, categorize, and investigate the effectiveness of passive cooling strategies (PCSs) for residential buildings. Forty-two studies published between 2000 and 2021 were reviewed; they examined the effects of PCSs on indoor temperature decrease, cooling load reduction, energy savings, and thermal comfort hour extension. In total, 30 passive strategies were identified and classified into three categories: design approach, building envelope, and passive cooling system. The review found that using various passive strategies can achieve, on average, (i) an indoor temperature decrease of 2.2 °C, (ii) a cooling load reduction of 31%, (iii) energy savings of 29%, and (v) a thermal comfort hour extension of 23%. Moreover, the five most effective passive strategies were identified as well as the differences between hot and dry climates and hot and humid climates.

Statistical shape modeling is the computational process of discovering significant shape parameters from segmented anatomies captured by medical images (such as MRI and CT scans), which can fully describe subject-specific anatomy in the context of a population. The presence of substantial non-linear variability in human anatomy often makes the traditional shape modeling process challenging. Deep learning techniques can learn complex non-linear representations of shapes and generate statistical shape models that are more faithful to the underlying population-level variability. However, existing deep learning models still have limitations and require established/optimized shape models for training. We propose Mesh2SSM, a new approach that leverages unsupervised, permutation-invariant representation learning to estimate how to deform a template point cloud to subject-specific meshes, forming a correspondence-based shape model. Mesh2SSM can also learn a population-specific template, reducing any bias due to template selection. The proposed method operates directly on meshes and is computationally efficient, making it an attractive alternative to traditional and deep learning-based SSM approaches.

Objective: We present the development of a non-contrast multi-parametric magnetic resonance (MPMR) imaging biomarker to assess treatment outcomes for magnetic resonance-guided focused ultrasound (MRgFUS) ablations of localized tumors. Images obtained immediately following MRgFUS ablation were inputs for voxel- wise supervised learning classifiers, trained using registered histology as a label for thermal necrosis. Methods: VX2 tumors in New Zealand white rabbits quadriceps were thermally ablated using an MRgFUS system under 3 T MRI guidance. Animals were re-imaged three days post-ablation and euthanized. Histological necrosis labels were created by 3D registration between MR images and digitized H&E segmentations of thermal necrosis to enable voxel- wise classification of necrosis. Supervised MPMR classifier inputs included maximum temperature rise, cumulative thermal dose (CTD), post-FUS differences in T2-weighted images, and apparent diffusion coefficient, or ADC, maps. A logistic regression, support vector machine, and random forest classifier were trained in red a leave-one-out strategy in test data from four subjects. Results: In the validation dataset, the MPMR classifiers achieved higher recall and Dice than than a clinically adopted 240 cumulative equivalent minutes at 43^∘ C (CEM ₄₃ ) threshold (0.43) in all subjects.redThe average Dice scores of overlap with the registered histological label for the logistic regression (0.63) and support vector machine (0.63) MPMR classifiers were within 6% of the acute contrast-enhanced non-perfused volume (0.67). Conclusions: Voxel- wise registration of MPMR data to histological outcomes facilitated supervised learning of an accurate non-contrast MR biomarker for MRgFUS ablations in a rabbit VX2 tumor model.

Background

The role of fiber orientation on a global chamber level in sustaining atrial fibrillation (AF) is unknown. The goal of this study was to correlate the fiber direction derived from Diffusion Tensor Imaging (DTI) with AF inducibility.

Transgenic goats with cardiac-specific overexpression of constitutively active TGF-β1 (n = 14) underwent AF inducibility testing by rapid pacing in the left atrium. We chose a minimum of 10 minutes of sustained AF as a cut-off for AF inducibility. Explanted hearts underwent DTI to determine the fiber direction. Using tractography data, we clustered, visualized, and quantified the fiber helix angles in 8 different regions of the left atrial wall using two reference vectors defined based on anatomical landmarks.

Sustained AF was induced in 7 out of 14 goats. The mean helix fiber angles in 7 out of 8 selected regions were statistically different (P-Value < 0.05) in the AF inducible group. The average fractional anisotropy (FA) and the mean diffusivity (MD) were similar in the two groups with FA of 0.32±0.08 and MD of 8.54±1.72 mm2/s in the non-inducible group and FA of 0.31±0.05 (P-value = 0.90) and MD of 8.68±1.60 mm2/s (P-value = 0.88) in the inducible group.

DTI based fiber direction shows significant variability across subjects with a significant difference between animals that are AF inducible versus animals that are not inducible. Fiber direction might be contributing to the initiation and sustaining of AF, and its role needs to be investigated further.

Statistical shape models (SSM) have been well-established as an excellent tool for identifying variations in the morphology of anatomy across the underlying population. Shape models use consistent shape representation across all the samples in a given cohort, which helps to compare shapes and identify the variations that can detect pathologies and help in formulating treatment plans. In medical imaging, computing these shape representations from CT/MRI scans requires time-intensive preprocessing operations, including but not limited to anatomy segmentation annotations, registration, and texture denoising. Deep learning models have demonstrated exceptional capabilities in learning shape representations directly from volumetric images, giving rise to highly effective and efficient Image-to-SSM. Nevertheless, these models are data-hungry and due to the limited availability of medical data, deep learning models tend to overfit. Offline data augmentation techniques, that use kernel density estimation based (KDE) methods for generating shape-augmented samples, have successfully aided Image-to-SSM networks in achieving comparable accuracy to traditional SSM methods. However, these augmentation methods focus on shape augmentation, whereas deep learning models exhibit image-based texture bias results in sub-optimal models. This paper introduces a novel strategy for on-the-fly data augmentation for the Image-to-SSM framework by leveraging data-dependent noise generation or texture augmentation. The proposed framework is trained as an adversary to the Image-to-SSM network, augmenting diverse and challenging noisy samples. Our approach achieves improved accuracy by encouraging the model to focus on the underlying geometry rather than relying solely on pixel values.

The microscopic evaluation of glands in the colon is of utmost importance in the diagnosis of inflammatory bowel disease (IBD) and cancer. When properly trained, deep learning pipelines can provide a systematic, reproducible, and quantitative assessment of disease-related changes in glandular tissue architecture. The training and testing of deep learning models require large amounts of manual annotations, which are difficult, time-consuming, and expensive to obtain. Here, we propose a method for the automated generation of ground truth in digital H&E slides using immunohistochemistry (IHC) labels. The image processing pipeline generates annotations of glands in H&E histopathology images from colon biopsies by transfer of gland masks from CK8/18, CDX2, or EpCAM IHC. The IHC gland outlines are transferred to co-registered H&E images for the training of deep learning models. We compare the performance of the deep learning models to manual annotations using an internal held-out set of biopsies as well as two public data sets. Our results show that EpCAM IHC provides gland outlines that closely match manual gland annotations (DICE = 0.89) and are robust to damage by inflammation. In addition, we propose a simple data sampling technique that allows models trained on data from several sources to be adapted to a new data source using just a few newly annotated samples. The best-performing models achieved average DICE scores of 0.902 and 0.89, respectively, on GLAS and CRAG colon cancer public datasets when trained with only 10% of annotated cases from either public cohort. Altogether, the performances of our models indicate that automated annotations using cell type-specific IHC markers can safely replace manual annotations. The automated IHC labels from single institution cohorts can be combined with small numbers of hand-annotated cases from multi-institutional cohorts to train models that generalize well to diverse data sources.

Semantic segmentation is a critical step in automated image interpretation and analysis where pixels are classified into one or more predefined semantically meaningful classes. Deep learning approaches for semantic segmentation rely on harnessing the power of annotated images to learn features indicative of these semantic classes. Nonetheless, they often fail to generalize when there is a significant domain (i.e., distributional) shift between the training (i.e., source) data and the dataset(s) encountered when deployed (i.e., target), necessitating manual annotations for the target data to achieve acceptable performance. This is especially important in medical imaging because different image modalities have significant intra- and inter-site variations due to protocol and vendor variability. Current techniques are sensitive to hyperparameter tuning and target dataset size. This paper presents an unsupervised domain adaptation approach for semantic segmentation that alleviates the need for annotating target data. Using kernel density estimation, we match the target data distribution to the source data in the feature space. We demonstrate that our results are comparable or superior on multiple-site prostate MRI and histopathology images, which mitigates the need for annotating target data.

Annotating medical imaging datasets is costly, so fine-tuning (or transfer learning) is the most effective method for digital pathology vision applications such as disease classification and semantic segmentation. However, due to texture bias in models trained on real-world images, transfer learning for histopathology applications might result in underperforming models, which necessitates the need for using unlabeled histopathology data and self-supervised methods to discover domain-specific characteristics. Here, we tested the premise that histopathology-specific pretrained models provide better initializations for pathology vision tasks, i.e., gland and cell segmentation. In this study, we compare the performance of gland and cell segmentation tasks with domain-specific and non-domain-specific pretrained weights. Moreover, we investigate the data size at which domain-specific pretraining produces a statistically significant difference in performance. In addition, we investigated whether domain-specific initialization improves the effectiveness of out-of-domain testing on distinct datasets but the same task. The results indicate that performance gain using domain-specific pretraining depends on both the task and the size of the training dataset. In instances with limited dataset sizes, a significant improvement in gland segmentation performance was also observed, whereas models trained on cell segmentation datasets exhibit no improvement.

In many scientific endeavors, increasingly abstract representations of data allow for new interpretive methodologies and conceptualization of phenomena. For example, moving from raw imaged pixels to segmented and reconstructed objects allows researchers new insights and means to direct their studies toward relevant areas. Thus, the development of new and improved methods for segmentation remains an active area of research. With advances in machine learning and neural networks, scientists have been focused on employing deep neural networks such as U-Net to obtain pixel-level segmentations, namely, defining associations between pixels and corresponding/referent objects and gathering those objects afterward. Topological analysis, such as the use of the Morse-Smale complex to encode regions of uniform gradient flow behavior, offers an alternative approach: first, create geometric priors, and then apply machine learning to classify. This approach is empirically motivated since phenomena of interest often appear as subsets of topological priors in many applications. Using topological elements not only reduces the learning space but also introduces the ability to use learnable geometries and connectivity to aid the classification of the segmentation target. In this paper, we describe an approach to creating learnable topological elements, explore the application of ML techniques to classification tasks in a number of areas, and demonstrate this approach as a viable alternative to pixel-level classification, with similar accuracy, improved execution time, and requiring marginal training data.

We present MedShapeNet, a large collection of anatomical shapes (e.g., bones, organs, vessels) and 3D surgical instrument models. Prior to the deep learning era, the broad application of statistical shape models (SSMs) in medical image analysis is evidence that shapes have been commonly used to describe medical data. Nowadays, however, state-of-the-art (SOTA) deep learning algorithms in medical imaging are predominantly voxel-based. In computer vision, on the contrary, shapes (including, voxel occupancy grids, meshes, point clouds and implicit surface models) are preferred data representations in 3D, as seen from the numerous shape-related publications in premier vision conferences, such as the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), as well as the increasing popularity of ShapeNet (about 51,300 models) and Princeton ModelNet (127,915 models) in computer vision research. MedShapeNet is created as an alternative to these commonly used shape benchmarks to facilitate the translation of data-driven vision algorithms to medical applications, and it extends the opportunities to adapt SOTA vision algorithms to solve critical medical problems. Besides, the majority of the medical shapes in MedShapeNet are modeled directly on the imaging data of real patients, and therefore it complements well existing shape benchmarks consisting of computer-aided design (CAD) models. MedShapeNet currently includes more than 100,000 medical shapes, and provides annotations in the form of paired data. It is therefore also a freely available repository of 3D models for extended reality (virtual reality - VR, augmented reality - AR, mixed reality - MR) and medical 3D printing. This white paper describes in detail the motivations behind MedShapeNet, the shape acquisition procedures, the use cases, as well as the usage of the online shape search portal: https://medshapenet.ikim.nrw/

Fourier Neural Operator (FNO) is a popular operator learning framework. It not only achieves the state-of-the-art performance in many tasks, but also is highly efficient in training and prediction. However, collecting training data for the FNO can be a costly bottleneck in practice, because it often demands expensive physical simulations. To overcome this problem, we propose Multi-Resolution Active learning of FNO (MRA-FNO), which can dynamically select the input functions and resolutions to lower the data cost as much as possible while optimizing the learning efficiency. Specifically, we propose a probabilistic multi-resolution FNO and use ensemble Monte-Carlo to develop an effective posterior inference algorithm. To conduct active learning, we maximize a utility-cost ratio as the acquisition function to acquire new examples and resolutions at each step. We use moment matching and the matrix determinant lemma to enable tractable, efficient utility computation. Furthermore, we develop a cost annealing framework to avoid over-penalizing high-resolution queries at the early stage. The over-penalization is severe when the cost difference is significant between the resolutions, which renders active learning often stuck at low-resolution queries and inferior performance. Our method overcomes this problem and applies to general multi-fidelity active learning and optimization problems. We have shown the advantage of our method in several benchmark operator learning tasks.

Neural network have achieved remarkable successes in many scientific fields. However, the interpretability of the neural network model is still a major bottlenecks to deploy such technique into our daily life. The challenge can dive into the non-linear behavior of the neural network, which rises a critical question that how a model use input feature to make a decision. The classical approach to address this challenge is feature attribution, which assigns an important score to each input feature and reveal its importance of current prediction. However, current feature attribution approaches often indicate the importance of each input feature without detail of how they are actually processed by a model internally. These attribution approaches often raise a concern that whether they highlight correct features for a model prediction.

For a neural network model, the non-linear behavior is often caused by non-linear activation units of a model. However, the computation behavior of a prediction from a neural network model is locally linear, because one prediction has only one activation pattern. Base on the observation, we propose an instance-wise linearization approach to reformulates the forward computation process of a neural network prediction. This approach reformulates different layers of convolution neural networks into linear matrix multiplication. Aggregating all layers' computation, a prediction complex convolution neural network operations can be described as a linear matrix multiplication F(x)=W⋅x+b. This equation can not only provides a feature attribution map that highlights the important of the input features but also tells how each input feature contributes to a prediction exactly. Furthermore, we discuss the application of this technique in both supervise classification and unsupervised neural network learning parametric t-SNE dimension reduction.