CVRG Projects

As the CVRG itself is composed of development teams at several institutions, the Project as a whole has been divided into 6 sub-projects. The first two projects focus on cardiovascular imaging and analysis. The third project focuses on the storage and analysis of electrocardiograms (ECGs). The fourth project focuses on clinical data management and analysis. The fifth project applies high-performance computing methods to the data transfer and analysis methods of the previous four projects. The sixth project reaches out to the cardiovascular community as a whole, transforming input from the community on the CVRG tools into paths for future development.

Project 1: Cardiac Computational Anatomy

(Director L Younes, with MI Miller and T Ratnanather)
The LDDMM algorithm registers an image/volume to a reference image/volume (atlas) by finding an optimal path between them. Populations of hearts can therefore be placed into a common coordinate system – that of the atlas. Analysis of the optimal path, based on the Euler-Lagrange equations that it satisfies, provides complete information on shape differences between the compared structures, expressed voxel by voxel. LDDMM, complemented by the PTDiff algorithm for motion analysis, therefore provide novel methods for the statistical analysis of shape and motion differences, expressed in a common coordinate system, at corresponding locations in populations of imaged hearts. These algorithms have been used to discover heart shape and motion biomarkers discriminating subjects with ischemic vs. non-ischemic cardiomyopathy with high statistical reliability. They have been applied to CT untagged and tagged MRI and ultrasound data. In this project, we will develop new data processing pipelines motivated by the needs of our DBPs, extend these methods to work with large image data sets, validate the methods in the DBPs, and deploy the tools on the CVRG. These data processing pipelines will be designed to use the DICOM image management and quality control capabilities of XNAT-CVI developed in Project 2.

Project 2: Cardiovascular Imaging Informatics

(CVII, Project co-Directors JJ Carr and Y. Ge)
Cardiovascular imaging (CVI) research requires quantitative, spatio-temporal imaging of the structure and function of the heart and vascular system. Further, co-registration of images with maps of electrical activity (Project 1) and strain, and analysis with other genotypic and phenotypic variables holds significant promise in linking basic science discoveries to clinical applications. To facilitate this translation, operational processes, data and analytic pipelines must be developed that address existing barriers to multi-site and interdisciplinary collaboration in CVI. In this project, we will develop and deploy a tool for managing quantitative, multidimensional CVI data. We will build on the eXtensible Neuroimaging Archive Toolkit (XNAT) an established part of the BIRN, to create the grid-enabled XNAT-CVI by extending core components of XNAT and adding modules specific for CVI. XNAT-CVI will become part of the CVRG data services infrastructure used in the DBPs shown in Table 1. XNAT-CVI will also be used in Project 1 for management of image analysis data and metadata. This work will therefore assure use of common tools across the CVRG imaging efforts. We will also develop grid-based methods for real-time reporting and tracking of imaging system performance during multisite studies. We will develop semantic tools for describing major imaging concepts in targeted DBPs with a special focus on cardiac strain data.

Project 3: ECG Data Management and Analysis

(Director RL Winslow, with S Granite)
Every clinical study of CV disease collects ECG data. Technology that makes it easier for clinician-researchers to interactively analyze and richly annotate ECG data using concepts from a well-designed ECG ontology will therefore have broad application and impact. It will bring tools for the analysis of ECG data into the hands of the medical experts who collect it, but who may lack technical skills. It will enable the dissemination and re-use of ECG data. It will enable integration of ECG data across studies in meta-analyses. This project will continue development of the CVRG infrastructure for managing and analyzing ECG data in order to deliver these capabilities.

Project 4: Integrative Clinical Information Management

(Director J Saltz; Investigators T Kurc, A Post)
Our objective is to develop and deploy software tools supporting the management and querying of clinical information, and the integration of clinical information with other multi-scale data collected in the DBPs. The OpenClinica data service currently deployed on the CVRG does not provide this capability. The ability to model clinical data and link it with semantically complex biomedical datasets benefits a wide range of studies. Examples include: a) systematically tracking groups of patients over time to explain risk factors for development and progression of disease; b) predicting and explaining patient response to treatment protocols by collecting and analyzing comprehensive sets of high throughput molecular data; and c) collecting different but interrelated data types through a closely coordinated set of experiments to answer a set of biomedical questions. Our effort will adapt and integrate existing tools, and undertake software development using service oriented architectures, and semantic web technologies.

Project 5: Analytical Workflows and Data Replication

(Director Ian Foster; Investigator Ravi Madduri)
Investigators in all of the DBPs need to perform complex data analyses involving multiple types of data and multiple steps of data processing. Currently, this is done by “manually” accessing data sets and running different analysis software one step at a time. In this aim, the Taverna workbench will be used to build workflows in which these data processing steps are automated, making it easier for non-experts to perform data analyses. DBP investigators also require that access to their data be highly reliable. To address this need, we will build and deploy CVRG data replication services to address this need.

Project 6: Management

(Director RL Winslow)
The CVRG will be governed in a way that addresses three concepts. Governance will be: a) transparent to assure the confidence of the CVRG user community and project teams; b) designed to assure that the needs of the CV community drive CVRG technology development; and c) designed to scale so that as the size of our user community grows, a process will be in place to accommodate that growth. The governance model will be a three-tiered structure comprised of the CVRG: a) Working Groups (WGs); b) Systems Integration Committee (SIC); and c) Executive Committee (EC).