Eliot L. Siegel, MD, FSIIM, University of Maryland School of Medicine, VA Maryland Health Care System

The caBIG™ (cancer Biomedical Informatics Grid) initiative of the National Cancer Institute is an information network enabling all constituencies in the cancer community to share data and knowledge to accelerate the discovery of new diagnostics and therapeutics, and to improve patient outcomes.[1,2] The purpose of this presentation is to provide an update of the activities of the National Cancer Imaging Archive (NCIA) and the caBIG™ Imaging Workspace as well as the caBIG™ initiative in general.

The NCIA effort has created a suite of software tools that provide a means of capturing, storing, and sharing medical images. Along with the RSNA’s clinical trials processing (CTP) software, its core functions include secure and de-identified acquisition and archival of medical images from clinical imaging modalities, support for federated storage, and queries utilizing specified parameters, role based retrieval of specified imaging studies for research review, and visualization of medical imaging datasets. The CTP software is free and open source and provides all of the processing capabilities in the MIRC clinical trials software with single-click installation, support for multiple pipelines with multiple configurable stages and support for multiple quarantines for data objects that are rejected during processing. The clinical trials processing software provides support for HTTP and DICOM import, a DICOM and XML anonymizer, file storage, database and HTTP and DICOM and FTP export. The caBIG™ Imaging workspace was created as a national multidisciplinary expert advisory board for the identification and prioritization of imaging informatics projects. The workspace has been in operation for more than three years and the projects within the workspace are in the process of being deployed in support of clinical trials and are being considered more broadly for commercial clinical applications as well.

Major projects include the creation of an ontology and model for DICOM and the use of this model to design a way to formulate and respond to semantic queries for information in the image and related information databases, the development of algorithm validation tools to create a way to measure the effectiveness of change detection and other software programs in medical imaging, as well as the development of a “reference” research workstation that can be used to retrieve, display, measure, annotate, and mark-up images stored on NCIA or associated with other research projects. Additionally, multiple practical implementations of the NCIA and imaging workspace software have been launched and will be briefly reviewed. The NCI has been working with other countries during the past year, such as the National Cancer Research Institute of the United Kingdom, to share research and development efforts as well as databases. The Vasari (visually assembled access to Rembrandt Images) project represents a cross cutting initiative to utilize data from the Rembrandt (Repository of Molecular Brain Neoplasia Data), which includes tumor specimen information including SNP array, expression array, proteomics, and clinical data from approximately 300 gliomas of the brain and correlates these with MRI images of the brain which have been annotated as well as patient survival information to enable a user to specify given structured imaging features and correlate them with the genetic and patient outcome data. We believe that this project will help attendees to understand the potential of a database which contains images, structured annotation, mark-ups, and reports to enable a personalized medicine approach in which imaging features can be more directly correlated with diagnostic possibilities, medical or radiation treatment options, and patient prognosis.

The NCIA and the caBIG™ imaging workspace have in addition to the international collaboration, worked together and with multiple research and development groups such as ACRIN, QARC, NTROI, the children’s oncology group and the radiation therapy oncology group in support of practical implementations of the software developed in association with workspace and NCIA projects. These efforts represent the largest organized imaging informatics effort to date with a strong emphasis on research in the cancer community and support for clinical trials but the work being done is very much applicable to the broader medical imaging community for clinical applications. These projects and initiatives will, in addition to the presentation, be showcased at the SIIM annual meeting with an emphasis on innovative and new practical applications. Overall, the NCIA and the caBIG™ imaging workspace have made excellent progress. The long term goal of the data repository and workspace is to broadly identify areas gaps in imaging informatics infrastructure such as the lack of a model of DICOM and lack of a standardized annotation and image mark-up system and also to bring advanced computing applications outside of the traditional medical space such as grid computing to our diagnostic imaging patients.

1. Harrington D. Imaging and Informatics at the National Cancer Institute, Part 2. Journal of the American College of Radiology. Vol.3(3):169–170.

2. Pan TC, Gurcan MN, Langella SA, et al. GridCAD: Grid-based Computer-aided Detection System. Radiographics. 2007;27:889-897.