PURPOSE:To develop a data-efficient strategy for accelerated MRI reconstruction with Diffusion Probabilistic Generative Models (DPMs) that enables faster scan times in clinical stroke MRI when only limited fully-sampled data are available. METHODS:Our simple training strategy first pre-trains a DPM on a large, diverse collection of publicly available fastMRI brain data and then fine-tunes on a small target dataset using carefully selected learning rates and fine-tuning durations. The approach is evaluated on controlled fastMRI experiments and on clinical stroke MRI data with a blinded clinical reader study. RESULTS:DPMs pre-trained on 4000 non-FLAIR subjects and fine-tuned on FLAIR data from only 20 target subjects achieve reconstruction performance comparable to models trained with substantially more target-domain FLAIR data across multiple acceleration factors. Moderate fine-tuning with a reduced learning rate yields improved performance, while insufficient or excessive fine-tuning degrades reconstruction quality. In a blinded reader study of 80 subjects at a single clinical site, images reconstructed from 2 × accelerated data using the proposed approach are rated comparably to standard-of-care on the image quality and structural delineation metrics defined in this work. CONCLUSION:Large-scale pre-training combined with targeted fine-tuning can enable DPM-based MRI reconstruction for our data-constrained, accelerated clinical stroke MRI application. In the single-site settings evaluated here, the proposed approach reduces the need for large application-specific datasets while maintaining clinically acceptable image quality, providing preliminary evidence for pre-trained and fine-tuned diffusion models as a strategy for accelerated MRI in targeted applications.

Recursive query computation, central to graph algorithms and relational databases, demands GPU acceleration due to its inherent computational intensity. While substantial prior work addresses GPU implementations of recursive queries that require fixed-point evaluation, existing systems are restricted to native execution environments. We introduce WGLog, the first web-browser-native GPU engine for compute-bound recursive database queries. WGLog is built entirely on WebGPU compute shaders, a cross-platform API that enables GPU acceleration in web browsers. WGLog leverages two key technical innovations. First, we replace hash-table-based joins with atomic-free sorted-array joins, eliminating the serialization bottleneck that hash tables suffer on skewed graphs. Second, we develop an asynchronous execution pipeline using WebGPU's indirect dispatch capability, which eliminates GPU-host synchronizations that would otherwise dominate per-iteration overhead. On representative workloads, WGLog delivers a 1.48–4.68x speedup over native GPU systems and orders-of-magnitude improvement over CPU and WebAssembly implementations.

Recent work has shown that implicit neural representations (INRs) can be trained to effectively compress structured and unstructured volume data, allowing for direct data querying with a reduced memory footprint. However, as existing INRs for unstructured volumes do not encode geometry, they require partial mesh storage for later sampling, limiting achievable compression. At the same time, novel view synthesis methods have shown that explicit collections of 3D Gaussians can be used to accurately visualize volume data. In this work, we introduce an explicit model for volume data compression based on 3D Gaussian primitives. We reinterpret collections of 3D Gaussians as an explicit representation of a scalar field and use a sampling strategy that reconstructs scalar values at spatial locations through weighted aggregation of intersecting Gaussians. We develop optimized CUDA-accelerated pipelines for structured and unstructured model sampling, loss functions that encourage accurate domain encoding by our models, and a novel sampling-error based densification strategy. Our explicit formulation naturally encodes domain geometry, eliminating the need for mesh storage in unstructured volumes and introducing significantly higher compression opportunities. Compared to existing INRs, we demonstrate that our explicit model achieves competitive reconstruction quality with significant training speedups on structured volumes, while markedly outperforming in all metrics on unstructured volumes.

Visualizing the large-scale datasets output by HPC resources presents a difficult challenge, as the memory and compute power required become prohibitively expensive for end user systems. Novel view synthesis techniques can address this by producing a small, interactive model of the data, requiring only a set of training images to learn from. While these models allow accessible visualization of large data and complex scenes, they do not provide the interactions needed for scientific volumes, as they do not support interactive selection of transfer functions and lighting parameters. To address this, we introduce Volume Encoding Gaussians (VEG), a 3D Gaussian-based representation for volume visualization that supports arbitrary color and opacity mappings. Unlike prior 3D Gaussian Splatting (3DGS) methods that store color and opacity for each Gaussian, VEG decouple the visual appearance from the data representation by encoding only scalar values, enabling transfer function-agnostic rendering of 3DGS models. To ensure complete scalar field coverage, we introduce an opacity-guided training strategy, using differentiable rendering with multiple transfer functions to optimize our data representation. This allows VEG to preserve fine features across a dataset's full scalar range while remaining independent of any specific transfer function. Across a diverse set of volume datasets, we demonstrate that our method outperforms the state-of-the-art on transfer functions unseen during training, while requiring a fraction of the memory and training time.

Datalog is a declarative logic-programming language used for complex analytic reasoning workloads such as program analysis and graph analytics. Datalog's popularity is due to its unique price-point, marrying logic-defined specification with the potential for massive data parallelism. While traditional engines are CPU-based, the memory-bound nature of Datalog has led to increasing interest in leveraging GPUs. These engines beat CPU-based engines by operationalizing iterated relational joins via SIMT-friendly join algorithms. Unfortunately, all existing GPU Datalog engines are built on binary joins, which are inadequate for the complex multi-way queries arising in production systems such as DOOP and ddisasm. For these queries, binary decomposition can incur the AGM bound asymptotic blowup in time and space, leading to OOM failures regardless of join order. Worst-Case Optimal Joins (WCOJ) avoid this blowup, but their attribute-at-a-time intersections map poorly to SIMT hardware under key skew, causing severe load imbalance across Streaming Multiprocessors (SMs). We present SRDatalog, the first GPU Datalog engine based on WCOJ. SRDatalog uses flat columnar storage and two-phase deterministic memory allocation to avoid the OOM failures of binary joins and the index-rebuild overheads of static WCOJ systems. To mitigate skew and hide hardware stalls, SRDatalog further employs root-level histogram-guided load balancing, structural helper-relation splitting, and stream-aligned rule multiplexing. On real-world program-analysis workloads, SRDatalog achieves geometric-mean speedups of 21x to 47x.

MPI_Allreduce is among the most performance-critical collectives in large-scale scientific computing and distributed machine learning, yet the small- and medium-message regime remains challenging: latency, synchronization depth, and strong hardware hierarchy between intra- and inter-domain communication all compound per-invocation cost. We present CHIARA, a configurable hierarchical Allreduce that encodes hardware hierarchy through a logical batch-lane topology and executes a staged schedule in which only a bounded portion of the reduction vector is active at a time. Inter-batch communication is distributed across multiple ranks via a rotating-root lane primitive, avoiding centralized leaders. Tool further enables a semi-composed Rabenseifner-style Allreduce by preserving a lane-aligned intermediate layout across the Reduce-Scatter/Allgather boundary, eliminating redundant intra-domain reorganization. We evaluate Tool on Polaris, Aurora, and Fugaku, achieving speedups of up to 1.94x, 13.43x, and 13.48x over vendor MPI_Allreduce, and up to 2.2x end-to-end speedup in a parallel k-means application.

Graph drawing, or graph layout creation, is a computationally difficult challenge in visualization that involves placing the vertices of a graph into a layout that provides insight into its structure. In order to visualize large-scale graphs, effective layouts are necessary for understanding. Previous work has shown the potential for graph drawing directly in the web browser by using WebGPU, a new API that brings the full capabilities of modern GPUs to the web. Compared to the existing state-of-the-art for web -based graph visualization, which rely on CPU -based graph drawing algorithms, WebGPU-accelerated work improves performance and scalability. However, we find that existing WebGPU solutions utilize suboptimal quadtree data structures for graph drawing. In this work, we implement a modified quadtree data structure that uses a Hilbert spatial ordering for a fully parallelizable bottom-up construction algorithm in WebGPU. We utilize this data structure, along with optimizations to the quadtree traversal, to propose a massively more performant graph drawing algorithm. We evaluate the performance of our work against the existing state-of-the-art and demonstrate up to 69.5 x speed-ups for layout creation of relevant graphs while enabling graph drawing for datasets of much larger size.

Datalog, a declarative logic programming language that operates bottom-up, has experienced increasing popularity due to its natural handling of recursive queries. Its applications span diverse fields, including graph mining, program analysis, deductive databases, and neuro-symbolic reasoning. While Datalog shares similarities with SQL in using relational algebra kernels, it uniquely employs iterative execution until reaching a fixed point to support recursion. Current Datalog engines like SLOG, LogicBlox, and Souffle work well with multi-core and multi-threaded systems, but none have yet tackled multi-node, multi-GPU architectures. Our research addresses this gap by developing the first multi-GPU, multinode Datalog engine. This advancement is particularly for high-performance computing (HPC) systems, which typically feature multiple GPUs per node. Our implementation combines MPI for inter-node communication with CUDA for GPU parallelization, enabling the processing of massive datasets in real time. We have created novel data-parallel implementations of core relational algebra operations ( join), while also optimizing deduplication and tuple materialization. To handle iterative execution, we have developed two novel GPU-accelerated methods for non-uniform all-to-all data exchange. Evaluating on Argonne National Lab's Polaris supercomputer demonstrated our engine's effectiveness, achieving performance improvements of up to 32x against state-of-the-art multi-node Datalog engine.

PURPOSE:To examine the effect of incorporating self-supervised denoising as a pre-processing step for training deep learning (DL) based reconstruction methods on data corrupted by Gaussian noise. K-space data employed for training are typically multi-coil and inherently noisy. Although DL-based reconstruction methods trained on fully sampled data can enable high reconstruction quality, obtaining large, noise-free datasets is impractical. METHODS:We leverage Generalized Stein's Unbiased Risk Estimate (GSURE) for denoising. We evaluate two DL-based reconstruction methods: Diffusion Probabilistic Models (DPMs) and Model-Based Deep Learning (MoDL). We evaluate the impact of denoising on the performance of these DL-based methods in solving accelerated multi-coil magnetic resonance imaging (MRI) reconstruction. The experiments were carried out on T2-weighted brain and fat-suppressed proton-density knee scans. RESULTS:We observed that self-supervised denoising enhances the quality and efficiency of MRI reconstructions across various scenarios. Specifically, employing denoised images rather than noisy counterparts when training DL networks results in lower normalized root mean squared error (NRMSE), higher structural similarity index measure (SSIM) and peak signal-to-noise ratio (PSNR) across different SNR levels, including 32, 22, and 12 dB for T2-weighted brain data, and 24, 14, and 4 dB for fat-suppressed knee data. CONCLUSION:We showed that denoising is an essential pre-processing technique capable of improving the efficacy of DL-based MRI reconstruction methods under diverse conditions. By refining the quality of input data, denoising enables training more effective DL networks, potentially bypassing the need for noise-free reference MRI scans.

MPI_Alltoallv generalizes the uniform all-to-all communication (MPI_Alltoall) by enabling the exchange of data-blocks of varied sizes among processes. This function plays a crucial role in facilitating many computational tasks, such as FFT calculations and graph mining operations. Popular MPI libraries, such as MPICH and OpenMPI, implement MPI_Alltoall using a combination of linear and logarithmic algorithms. However, MPI_Alltoallv typically relies only on variations of linear algorithms, missing the benefits of logarithmic approaches. Furthermore, current algorithms also overlook the intricacies of modern HPC system architectures, such as the significant performance gap between intra-node (local) and inter-node (global) communication. To address these problems, this paper presents two novel algorithms: Parameterized Logarithmic non-uniform All-to-all (ParLogNa) and Parameterized Linear nonuniform All-to-all (ParLinNa). ParLogNa is a tunable logarithmic time algorithm for non-uniform all-to-all, and ParLinNa is a hierarchical and tunable near-linear-time algorithm for non-uniform all-to-all. These algorithms efficiently address the trade-off between bandwidth maximization and latency minimization that existing implementations struggle to optimize. We show a performance improvement over the state-of-the-art implementations by factors of 42x and 138x on Polaris and Fugaku, respectively.

Modern Datalog engines (e.g., LogicBlox, Souffle, ddlog) enable their users to write declarative queries which compute recursive deductions over extensional facts, leaving high-performance operationalization (query planning, seminaive evaluation, and parallelization) to the engine. Such engines form the backbone of modern high-throughput applications in static analysis, network monitoring, and socialmedia mining. In this paper, we present a methodology for implementing a modern in-memory Datalog engine on data center GPUs, allowing us to achieve significant (up to 45x) gains compared to Souffle (a modern CPU-based engine) on context-sensitive points-to analysis of PostgreSQL. We present GPUlog, a Datalog engine backend that implements iterated relational algebra kernels over a novel rangeindexed data structure we call the hash-indexed sorted array (HISA). HISA combines the algorithmic benefits of incremental range-indexed relations with the raw computation throughput of operations over dense data structures. Our experiments show that GPUlog is significantly faster than CPU-based Datalog engines while achieving a favorable memory footprint compared to contemporary GPU-based joins.

Purpose To combine deep learning and biology-based modeling to predict the response of locally advanced, triple-negative breast cancer before initiating neoadjuvant chemotherapy (NAC). Materials and Methods In this retrospective study, a biology-based mathematical model of tumor response to NAC was constructed and calibrated on a patient-specific basis using imaging data from patients enrolled in the MD Anderson A Robust TNBC Evaluation FraMework to Improve Survival trial (ARTEMIS; ClinicalTrials.gov registration no. NCT02276443) between April 2018 and May 2021. To relate the calibrated parameters in the biology-based model and pretreatment MRI data, a convolutional neural network (CNN) was employed. The CNN predictions of the calibrated model parameters were used to estimate tumor response at the end of NAC. CNN performance in the estimations of total tumor volume (TTV), total tumor cellularity (TTC), and tumor status was evaluated. Model-predicted TTC and TTV measurements were compared with MRI-based measurements using the concordance correlation coefficient and area under the receiver operating characteristic curve (for predicting pathologic complete response at the end of NAC). Results The study included 118 female patients (median age, 51 years [range, 29-78 years]). For comparison of CNN predicted to measured change in TTC and TTV over the course of NAC, the concordance correlation coefficient values were 0.95 (95% CI: 0.90, 0.98) and 0.94 (95% CI: 0.87, 0.97), respectively. CNN-predicted TTC and TTV had an area under the receiver operating characteristic curve of 0.72 (95% CI: 0.34, 0.94) and 0.72 (95% CI: 0.40, 0.95) for predicting tumor status at the time of surgery, respectively. Conclusion Deep learning integrated with a biology-based mathematical model showed good performance in predicting the spatial and temporal evolution of a patient's tumor during NAC using only pre-NAC MRI data. Keywords: Triple-Negative Breast Cancer, Neoadjuvant Chemotherapy, Convolutional Neural Network, Biology-based Mathematical Model Supplemental material is available for this article. Clinical trial registration no. NCT02276443 ©RSNA, 2024 See also commentary by Mei and Huang in this issue.

Datalog is a logic programming language widely used in knowledge representation and reasoning (KRR), program analysis, and social media mining due to its expressiveness and high performance. Traditionally, Datalog engines use either row-oriented or column-oriented storage. Engines like VLog and Nemo favor column-oriented storage for efficiency on limited-resource machines, while row-oriented engines like Soufflé use advanced datastructures with locking to perform better on multi-core CPUs. The advent of modern datacenter GPUs, such as the NVIDIA H100 with its ability to run over 16k threads simultaneously and high memory bandwidth, has reopened the debate on which storage layout is more effective. This paper presents the first column-oriented Datalog engines tailored to the strengths of modern GPUs. We present VFLog, a CUDA-based Datalog runtime library with a column-oriented GPU datastructure that supports all necessary relational algebra operations. Our results demonstrate over 200x performance gains over SOTA CPU-based column-oriented Datalog engines and a 2.5x speedup over GPU Datalog engines in various workloads, including KRR.

Mapping the extent of flood events is a necessary and important aspect of disaster management. In recent years, deep learning methods have evolved as an effective tool to quickly label high-resolution imagery and provide necessary flood extent mappings. These methods, though, require large amounts of annotated training data to create models that are accurate and robust to new flooded imagery. In this work, we present FloodTrace, a web -based application that enables effective crowdsourcing of flooded region annotation for machine learning applications. To create this application, we conducted extensive interviews with domain experts to produce a set of formal requirements. Our work brings topological segmentation tools to the web and greatly improves annotation efficiency compared to the state-of-the-art. The user-friendliness of our solution allows researchers to outsource annotations to non-experts and utilize them to produce training data with equal quality to fully expert-labeled data. We conducted a user study to confirm our application's effectiveness in which 266 graduate students annotated high-resolution aerial imagery from Hurricane Matthew in North Carolina. Experimental results show the efficiency benefits of our application for untrained users, with median annotation time less than half the state-of-the-art annotation method. In addition, using our application's aggregation and correction framework, flood detection models trained on crowdsourced annotations were able to achieve performance equal to models trained on fully expert-labeled annotations, while requiring a fraction of the expert's time.

Cancer is a complex disease affecting various organs and is a major cause of death worldwide. During cancer initiation, disease progression, and tumor metastasis, various genomic and proteomic alterations are observed. Recent technological advances have led to the generation of large amounts of molecular data, including genomics and transcriptomics. These large-scale datasets can be utilized to analyze and identify sub-class-specific cancer biomarkers and targets. However, there is a need for the development of user-friendly tools for large-scale data analysis, disseminating the analyzed data in a visualizable format to cancer researchers with no programming skills. We developed UALCAN, a comprehensive platform that allows users to integrate disparate data to better understand the genes, proteins, and pathways perturbed in cancer and make discoveries of potential biomarkers and targets. In the current study, we describe the development of the UALCAN Mobile application (app) that will provide cancer transcriptomic data obtained from The Cancer Genome Atlas (TCGA) project to evaluate protein-coding gene expression based on various stratifications, including stage, grade, race, gender, and molecular-subtypes across over 30 types of cancers. In addition, the UALCAN mobile provides data analysis options for epigenetic changes due to DNA promoter methylation and Clinical Proteomic Tumor Analysis Consortium (CPTAC) cancer proteomic data. The app provides access to large cancer molecular datasets on the go. To find changes in the expression of causative genes and proteins and to identify biomarkers and therapeutic targets, UALCAN mobile app will be extremely valuable. The "UALCAN Mobile" app is free to use and can be downloaded from both the iOS/Apple and the Android Play Store and has been downloaded over 100 times in each of iOS and android app stores.

Using temporal analysis of fMRI (functional Magnetic Resonance Imaging) data, we can characterize dynamic changes in brain connectivity over time. However, dynamic temporal analysis of fMRI data is challenging due to the high dimensionality of the datasets. Another fundamental challenge of dynamic temporal analysis of fMRI is the presence of non-neural artifacts that add sources of variation in the data that are not directly related to brain activity. For example, when data are acquired at different scanners at different temporal sampling rates and later analyzed as a single dataset, we have to contend with different number of image snapshots for different subjects. Also, high-frequency scans lead to more fine-grained temporal snapshotting than low-frequency scans. These factors can obscure true neural signals and lead to inconsistent characterization of dynamic brain connectivity across scans. Existing graph-based solutions often struggle with parameter sensitivity, since their outcomes depend heavily on selecting an arbitrary correlation threshold for defining network edges. In contrast, topological data analysis (TDA) sweeps across all threshold values to track the persistence of connectivity features, making it more robust for capturing fine-grained temporal dynamics. Clustering methods become imperative in this context as they offer a powerful means to uncover underlying structures within the high-dimensional temporal data. We address these challenges by developing a topological data analysis based temporal clustering pipeline targeted for dynamic functional connectivity derived from fMRI datasets that can preserve the dynamics of the temporal datasets and mask out the non-neural variability induced by varying sampling rates. The TDA-based pipeline extracts robust features that are invariant to non-neural noise and uses them to perform temporal clustering. We evaluate our framework by performing temporal clustering of resting-state fMRI-derived dynamic functional connectivity brain networks obtained from 316 subjects, each of whom was scanned thrice using different temporal sampling periods. The efficacy of our TDA-based pipeline is compared against three alternative approaches: direct time-series clustering, PCA-based dimensionality reduction and clustering, and a traditional fully connected network analysis pipeline with MDS-based dimensionality reduction. Additionally, we demonstrate that for a majority of cases, the number of clusters remains consistent for the same subjects scanned at different temporal sampling rates-showcasing the greater robustness of our TDA-based pipeline compared to other pipelines. The TDA pipeline achieved higher overlaps (59 %) in optimal cluster numbers across sampling cohorts, as well as higher pairwise similarity (74-77 %) between subjects' cluster solutions. This indicates that incorporating network topology via TDA enables more robust clustering of temporal fMRI datasets despite changes in sampling rates. Furthermore, we validate our method on a clinical dataset (ADHD-200). The TDA-based pipeline successfully captures consistent clustering patterns across different sites and scanning protocols, with higher stability of cluster assignments (> 80% similarity) and better separation of subject-level dynamics compared to existing approaches. This reinforces the method's robustness in multi-site, multi-condition settings. Our results demonstrate that incorporating network topology via TDA significantly enhances the reliability of temporal clustering in fMRI studies, offering a robust framework for studying brain dynamics across heterogeneous acquisition settings.

We provide a framework for solving inverse problems with diffusion models learned from linearly corrupted data. Firstly, we extend the Ambient Diffusion framework to enable training directly from measurements corrupted in the Fourier domain. Subsequently, we train diffusion models for MRI with access only to Fourier subsampled multi-coil measurements at acceleration factors R= 2,4,6,8. Secondly, we propose Ambient Diffusion Posterior Sampling (A-DPS), a reconstruction algorithm that leverages generative models pre-trained on one type of corruption (e.g. image inpainting) to perform posterior sampling on measurements from a different forward process (e.g. image blurring). For MRI reconstruction in high acceleration regimes, we observe that A-DPS models trained on subsampled data are better suited to solving inverse problems than models trained on fully sampled data. We also test the efficacy of A-DPS on natural image datasets (CelebA, FFHQ, and AFHQ) and show that A-DPS can sometimes outperform models trained on clean data for several image restoration tasks in both speed and performance.

Background: Patients with advanced cancer and their family caregivers often experience poor quality of life (QOL). Self- report measures are commonly used to quantify QOL of family caregivers but may have limitations such as recall bias and social desirability bias. Variables derived from passively obtained smartphone GPS data are a novel approach to measuring QOL that may overcome these limitations and enable detection of early signs of mental and physical health (PH) deterioration. Objective: This study explored the feasibility of a digital phenotyping approach by assessing participant adherence and examining correlations between smartphone GPS data and QOL levels among family caregivers and patients with advanced cancer. Methods: This was a secondary analysis involving 7 family caregivers and 4 patients with advanced cancer that assessed correlations between GPS sensor data captured by a personally owned smartphone and QOL self-report measures over 12 weeks through linear correlation coefficients. QOL as measured by the Patient-Reported Outcomes Measurement Information System (PROMIS) Global Health 10 was collected at baseline, 6, and 12 weeks. Using a Beiwe smartphone app, GPS data were collected and processed into variables including total distance, time spent at home, transition time, and number of significant locations. Results: The study identified relevant temporal correlations between QOL and smartphone GPS data across specific time periods. For instance, in terms of PH, associations were observed with the total distance traveled (12 and 13 wk, with r ranging 0.37 to 0.38), time spent at home (-4 to -2 wk, with r ranging from -0.41 to -0.49), and transition time (-4 to -2 wk, with r ranging -0.38 to -0.47). Conclusions: This research offers insights into using passively obtained smartphone GPS data as a novel approach for assessing and monitoring QOL among family caregivers and patients with advanced cancer, presenting potential advantages over traditional self-report measures. The observed correlations underscore the potential of this method to detect early signs of deteriorating mental health and PH, providing opportunities for timely intervention and support.

Non-uniform all-to-all communication patterns present optimization challenges in parallel computing due to their irregular data distribution and dynamic behavior. While MPI_Alltoallv provides the standard interface for such exchanges, achieving optimal performance requires careful selection among multiple implementation variants and tuning of algorithm-specific parameters. This paper presents a data-driven autotuning framework that combines machine learning-based runtime prediction with a lookup-table mechanism for fast configuration selection. The ML model estimates the communication time of each algorithm configuration under a given system setup, allowing the framework to identify the optimal implementation and parameter set based on predicted performance. We validate our approach through comprehensive benchmarking of MPI_Alltoallv and two specialized algorithms across varying process counts, message sizes, and tunable parameters. Applied to a real MPI-based transitive closure application on the Fugaku supercomputer, our framework achieves up to 6.03x reduction in communication time over the vendor implementation, providing a detailed understanding of non-uniform collective communication behavior and a practical framework for automatic performance optimization in HPC applications.