OBJECTIVE: To develop ImageVU, a scalable research imaging infrastructure that integrates clinical imaging data with metadata-driven cohort discovery, enabling secure, efficient, and regulatory-compliant access to imaging for secondary and opportunistic research use. This manuscript presents a detailed description of ImageVU's key components and lessons learned to assist other institutions in developing similar research imaging services and infrastructure. METHODS: ImageVU was designed to support the secondary use of radiological imaging data through a dedicated research imaging store. The system comprises four interconnected components: a Research PACS, an Ad Hoc Backfill Host, Cloud Storage System, and a De-Identification System. Imaging metadata are extracted and stored in the Research Derivative (RD), an identified clinical data repository, and the Synthetic Derivative (SD), a de-identified research data repository, with access facilitated through the RD Discover web portal. Researchers interact with the system via structured metadata queries and multiple data delivery options, including web-based viewing, bulk downloads, and dataset preparation for high-performance computing environments. RESULTS: The integration of metadata-driven search capabilities has streamlined cohort discovery and improved imaging data accessibility. As of December 2024, ImageVU has processed 12.9 million MRI and CT series from 1.36 million studies across 453,403 patients. The system has supported 75 project requests, delivering over 50 TB of imaging data to 55 investigators, leading to 66 published research papers. CONCLUSION: ImageVU demonstrates a scalable and efficient approach for integrating clinical imaging into research workflows. By combining institutional data infrastructure with cloud-based storage and metadata-driven cohort identification, the platform enables secure and compliant access to imaging for translational research.

Analyzing high resolution whole slide images (WSIs) with regard to information across multiple scales poses a significant challenge in digital pathology. Multi-instance learning (MIL) is a common solution for working with high resolution images by classifying bags of objects (i.e. sets of smaller image patches). However, such processing is typically performed at a single scale (e.g., 20× magnification) of WSIs, disregarding the vital inter-scale information that is key to diagnoses by human pathologists. In this study, we propose a novel cross-scale MIL algorithm to explicitly aggregate inter-scale relationships into a single MIL network for pathological image diagnosis. The contribution of this paper is three-fold: (1) A novel cross-scale MIL (CS-MIL) algorithm that integrates the multi-scale information and the inter-scale relationships is proposed; (2) A toy dataset with scale-specific morphological features is created and released to examine and visualize differential cross-scale attention; (3) Superior performance on both in-house and public datasets is demonstrated by our simple cross-scale MIL strategy. The official implementation is publicly available at https://github.com/hrlblab/CS-MIL.

Medical image segmentation, or computing voxel-wise semantic masks, is a fundamental yet challenging task in medical imaging domain. To increase the ability of encoder-decoder neural networks to perform this task across large clinical cohorts, contrastive learning provides an opportunity to stabilize model initialization and enhances downstream tasks performance without ground-truth voxel-wise labels. However, multiple target objects with different semantic meanings and contrast level may exist in a single image, which poses a problem for adapting traditional contrastive learning methods from prevalent "image-level classification" to "pixel-level segmentation". In this article, we propose a simple semantic-aware contrastive learning approach leveraging attention masks and image-wise labels to advance multi-object semantic segmentation. Briefly, we embed different semantic objects to different clusters rather than the traditional image-level embeddings. We evaluate our proposed method on a multi-organ medical image segmentation task with both in-house data and MICCAI Challenge 2015 BTCV datasets. Compared with current state-of-the-art training strategies, our proposed pipeline yields a substantial improvement of 5.53% and 6.09% on Dice score for both medical image segmentation cohorts respectively (p-value 0.01). The performance of the proposed method is further assessed on external medical image cohort via MICCAI Challenge FLARE 2021 dataset, and achieves a substantial improvement from Dice 0.922 to 0.933 (p-value 0.01).

Although machine learning (ML)-based models are increasingly being used by cloud-based data-driven services, two key problems exist when used at the edge. First, the size and complexity of these models hampers their deployment at the edge, where heterogeneity of resource types and constraints on resources is the norm. Second, ML models are known to be vulnerable to adversarial perturbations. To address the edge deployment issue, model compression techniques, especially model quantization, have shown significant promise. However, the adversarial robustness of such quantized models remains mostly an open problem. To address this challenge, this paper investigates whether quantized models with different precision levels can be vulnerable to the same universal adversarial perturbation (UAP). Based on these insights, the paper then presents a cloud-native service that generates and distributes adversarially robust compressed models deployable at the edge using a novel, defensive post-training quantization approach. Experimental evaluations reveal that although quantized models are vulnerable to UAPs, post-training quantization on the synthesized, adversarially-trained models are effective against such UAPs. Furthermore, deployments on heterogeneous edge devices with flexible quantization settings are efficient thereby paving the way in realizing adversarially robust data-driven cloud/edge services.

Body part regression is a promising new technique that enables content navigation through self-supervised learning. Using this technique, the global quantitative spatial location for each axial view slice is obtained from computed tomography (CT). However, it is challenging to define a unified global coordinate system for body CT scans due to the large variabilities in image resolution, contrasts, sequences, and patient anatomy. Therefore, the widely used supervised learning approach cannot be easily deployed. To address these concerns, we propose an annotation-free method named blind-unsupervised-supervision network (BUSN). The contributions of the work are in four folds: (1) 1030 multi-center CT scans are used in developing BUSN without any manual annotation. (2) the proposed BUSN corrects the predictions from unsupervised learning and uses the corrected results as the new supervision; (3) to improve the consistency of predictions, we propose a novel neighbor message passing (NMP) scheme that is integrated with BUSN as a statistical learning based correction; and (4) we introduce a new pre-processing pipeline with inclusion of the BUSN, which is validated on 3D multi-organ segmentation. The proposed method is trained on 1,030 whole body CT scans (230,650 slices) from five datasets, as well as an independent external validation cohort with 100 scans. From the body part regression results, the proposed BUSN achieved significantly higher median R-squared score (=0.9089) than the state-of-the-art unsupervised method (=0.7153). When introducing BUSN as a preprocessing stage in volumetric segmentation, the proposed pre-processing pipeline using BUSN approach increases the total mean Dice score of the 3D abdominal multi-organ segmentation from 0.7991 to 0.8145.

Smart Internet of Things (IoT) applications require real-time and robust predictive analytics, which are based on Machine Learning (ML) models. Building ML models from Big Data is not only time-consuming, but developers often lack the needed expertise for feature engineering, parameter tuning, and model selection. The proliferation of ML libraries and frameworks, data ingestion tools, stream and batch processing engines, visualization techniques, and the range of available hardware platforms further exacerbates the system design, rapid development, and deployment problems. Finally, resource constraints of IoT require that the execution of the analytics engine be distributed across the cloud-edge spectrum. To overcome these daunting challenges, we present Stratum, which is an event-driven Big Data-as-a-Service offering for IoT analytics lifecycle management. Stratum provides users with an intuitive, declarative mechanism based on the principles of model-driven engineering to specify the application and infrastructure requirements. It automates the deployment via generative programming principles. This paper highlights the problems that Stratum resolves, demonstrating its capabilities using real-world case studies.

The findings of splenomegaly, abnormal enlargement of the spleen, is a non-invasive clinical biomarker for liver and spleen diseases. Automated segmentation methods are essential to efficiently quantify splenomegaly from clinically acquired abdominal magnetic resonance imaging (MRI) scans. However, the task is challenging due to: 1) large anatomical and spatial variations of splenomegaly; 2) large inter- and intra-scan intensity variations on multi-modal MRI; and 3) limited numbers of labeled splenomegaly scans. In this paper, we propose the Splenomegaly Segmentation Network (SS-Net) to introduce the deep convolutional neural network (DCNN) approaches in multi-modal MRI splenomegaly segmentation. Large convolutional kernel layers were used to address the spatial and anatomical variations, while the conditional generative adversarial networks were employed to leverage the segmentation performance of SS-Net in an end-to-end manner. A clinically acquired cohort containing both T1-weighted (T1w) and T2-weighted (T2w) MRI splenomegaly scans was used to train and evaluate the performance of multi-atlas segmentation (MAS), 2D DCNN networks, and a 3-D DCNN network. From the experimental results, the DCNN methods achieved superior performance to the state-of-the-art MAS method. The proposed SS-Net method has achieved the highest median and mean Dice scores among the investigated baseline DCNN methods.

A key limitation of deep convolutional neural networks (DCNN) based image segmentation methods is the lack of generalizability. Manually traced training images are typically required when segmenting organs in a new imaging modality or from distinct disease cohort. The manual efforts can be alleviated if the manually traced images in one imaging modality (e.g., MRI) are able to train a segmentation network for another imaging modality (e.g., CT). In this paper, we propose an end-to-end synthetic segmentation network (SynSeg-Net) to train a segmentation network for a target imaging modality without having manual labels. SynSeg-Net is trained by using (1) unpaired intensity images from source and target modalities, and (2) manual labels only from source modality. SynSeg-Net is enabled by the recent advances of cycle generative adversarial networks (CycleGAN) and DCNN. We evaluate the performance of the SynSeg-Net on two experiments: (1) MRI to CT splenomegaly synthetic segmentation for abdominal images, and (2) CT to MRI total intracranial volume synthetic segmentation (TICV) for brain images. The proposed end-to-end approach achieved superior performance to two stage methods. Moreover, the SynSeg-Net achieved comparable performance to the traditional segmentation network using target modality labels in certain scenarios. The source code of SynSeg-Net is publicly available 2.

Big data medical image processing applications involving multi-stage analysis often exhibit significant variability in processing times ranging from a few seconds to several days. Moreover, due to the sequential nature of executing the analysis stages enforced by traditional software technologies and platforms, any errors in the pipeline are only detected at the later stages despite the sources of errors predominantly being the highly compute-intensive first stage. This wastes precious computing resources and incurs prohibitively higher costs for re-executing the application. The medical image processing community to date remains largely unaware of these issues and continues to use traditional high-performance computing clusters, which incur a high operating cost due to the use of dedicated resources and expensive centralized file systems. To overcome these challenges, this paper proposes an alternative approach for multi-stage analysis in medical image processing by using the Apache Hadoop ecosystem and offering it as a service in the cloud. We make the following contributions. First, we propose a concurrent pipeline execution framework and an associated semi-automatic, real-time monitoring and checkpointing framework that can detect outliers and achieve quality assurance without having to completely execute the expensive first stage of processing thereby expediting the entire multi-stage analysis. Second, we present a simulator to rapidly estimate the execution time for a given multi-stage analysis, which can aid the users in deciding the appropriate approach for their use cases. We conduct empirical evaluation of our framework and show that it requires 76.75% lesser wall time and 29.22% lesser resource time compared to the traditional approach that lacks such a quality assurance mechanism.

Large-scale medical imaging studies to date have predominantly leveraged in-house, laboratory-based or traditional grid computing resources for their computing needs, where the applications often use hierarchical data structures (e.g., NFS file stores) or databases (e.g., COINS, XNAT) for storage and retrieval. The resulting performance for laboratory-base approaches reveal that performance is impeded by standard network switches since they can saturate network bandwidth during transfer from storage to processing nodes for even moderate-sized studies. On the other hand, the grid may be costly to use due to the dedicated resources used to execute the tasks and lack of elasticity. With increasing availability of cloud-based Big Data frameworks, such as Apache Hadoop, cloud-based services for executing medical imaging studies have shown promise. Despite this promise, our preliminary studies have revealed that existing Big Data frameworks illustrate different performance limitations for medical imaging applications, which calls for new algorithms that optimize their performance and suitability for medical imaging. For instance, Apache HBase's load distribution strategy of region split and merge is detrimental to the hierarchical organization of imaging data (e.g., project, subject, session, scan, slice). To address these challenges, this doctoral research is developing a range of performance optimization algorithms. This paper describes preliminary research we have conducted in this realm and presents a list of research tasks that will be undertaken as part of this doctoral research.

Traditional in-house, laboratory-based medical imaging studies use hierarchical data structures (e.g., NFS file stores) or databases (e.g., COINS, XNAT) for storage and retrieval. The resulting performance from these approaches is, however, impeded by standard network switches since they can saturate network bandwidth during transfer from storage to processing nodes for even moderate-sized studies. To that end, a cloud-based "medical image processing-as-a-service" offers promise in utilizing the ecosystem of Apache Hadoop, which is a flexible framework providing distributed, scalable, fault tolerant storage and parallel computational modules, and HBase, which is a NoSQL database built atop Hadoop's distributed file system. Despite this promise, HBase's load distribution strategy of region split and merge is detrimental to the hierarchical organization of imaging data (e.g., project, subject, session, scan, slice). This paper makes two contributions to address these concerns by describing key cloud engineering principles and technology enhancements we made to the Apache Hadoop ecosystem for medical imaging applications. First, we propose a row-key design for HBase, which is a necessary step that is driven by the hierarchical organization of imaging data. Second, we propose a novel data allocation policy within HBase to strongly enforce collocation of hierarchically related imaging data. The proposed enhancements accelerate data processing by minimizing network usage and localizing processing to machines where the data already exist. Moreover, our approach is amenable to the traditional scan, subject, and project-level analysis procedures, and is compatible with standard command line/scriptable image processing software. Experimental results for an illustrative sample of imaging data reveals that our new HBase policy results in a three-fold time improvement in conversion of classic DICOM to NiFTI file formats when compared with the default HBase region split policy, and nearly a six-fold improvement over a commonly available network file system (NFS) approach even for relatively small file sets. Moreover, file access latency is lower than network attached storage.

With developing an affordable, easily accessible and scalable online CPS laboratory to promote CPS education system, we are faced with and focused on a number of cyber-physical challenges including the model design and simulation strategies. The authors provide a complete process to simulate a behavior of a user-design CPS conveyor system. The user-design model is sent to the background, treated offline, and extracted the simulation result and finally feedback to user as an animation. The solution approach has two main parts, as the aspect of the modeling work, complex domain-specific conveyor design are defined in the Generic Modeling Environment (GME), it can be mapped and transformed to the global grid, another domain-specific model, which contains only one kind of node with huge dimension so that all different species of components in complex model are mapped to the typical nodes in grid, and it is easy to operate and simulate the nodes in global grid to fit for the need when multiple experiments being mapped to the grid. In this work, we only concerned the scenario of one experiment. The transformation and mapping process is implemented through Graph Rewriting and Transformation. As a background simulation, the Robocodes code is automatically generated by GME interpreter from global grid and is applied to generate the path logic to transmit the package, according to the package type in each input ports. After acquiring the transmit speed and path, Robocode simulation outputs the coordinate and time information to generate the Java animation. The final Java animation will be feedback to the user side to see the result of package transmission flow.