The National Cancer Imaging Archive and caIMAGE: Enabling Standardization and Collaboration
Imaging technologies—such as computed tomography (CT), magnetic resonance imaging (MRI), and positron-emission tomography (PET)—play a vital role in better detection, diagnosis, staging, and grading of cancer. More than 300 million medical imaging procedures are performed each year in the United States alone, and the demand for, and dependence on, diagnostic imaging for cancer research and treatment has exploded over the last few years.
“Every patient who has cancer now has a fairly extensive imaging work-up for that cancer,” said Eliot L. Siegel, M.D., Vice Chair of Information Systems in the University of Maryland School of Medicine’s Department of Diagnostic Radiology and Nuclear Medicine, and Chief of Imaging, Veterans Affairs Maryland Healthcare System. Dr. Siegel also leads the caBIG™ In Vivo Imaging Workspace and has responsibility for the National Cancer Imaging Archive (NCIA).
Increasing Collaboration to Improve Patient Outcomes
Greater reliance on imaging in clinical care and biomedical research, the current structure of clinical studies, and other advances in technology have led to an enormous increase in the quantity and complexity of cancer imaging data. Despite this increase, however, there is little standardization of protocols used to acquire, analyze, and annotate images among cancer centers and other research facilities.
“Advancements in computer technology, and consequently in the complexity and sophistication of imaging studies, have greatly increased our ability to screen high-risk patients, diagnose patients suspected of having the disease, and then stage patients prior to surgery at individual centers. However, in order to truly advance this field, we need to standardize the way that we access, analyze, annotate and share these images,” said Dr. Siegel.
This current lack of standards for collaboration and analysis means that clinical providers and researchers must continue to rely on the small, limited sets of images available in their own institutional databases when coordinating treatment plans or drawing clinical and research conclusions about cancer or on the general literature. In order to overcome this disconnect, caBIG™ has introduced a number of applications through its In Vivo Imaging Workspace.
One of the primary goals of the caBIG™ imaging program is to connect the wide variety of imaging databases that exist across cancer centers, research labs, and hospitals. Better standardization and collaboration around image collection and analysis is expected to:
- Produce a standard for a global library of databases that serve as a platform for the sharing, optimization, and effective integration of all collected imaging information into cancer research;
- Facilitate computer-aided diagnoses and hypothesis-driven research among scientists from government, academia, and the private sector;
- Equip clinical providers and researchers with larger, more complex and representative sets of imaging data to use in identifying and drawing conclusions for research and treatment; and
- Expedite the analysis of images across multiple disparate databases through the use of standardized imaging algorithms and annotation methods.
Fostering a Personalized Approach
Two of these applications, the National Cancer Imaging Archive (NCIA) (which currently focuses on diagnostic imaging) and caIMAGE (which focuses on pathology), are free, open-source tools that have been built to help streamline the imaging sharing process for the benefit of both clinical research and patients. These tools were specifically designed to foster research collaboration and, ultimately, a more personalized approach to treatment.
“One of our major efforts is to create standards for how images are annotated and how they are marked up, and to provide meaning to those images in a way that could either be read by humans or computer databases,” said Dr. Siegel. “This effort gets to the heart of the caBIG™ initiative, which is to facilitate the transition to stratified or personalized medicine and a semantic ‘world wide web’ of cancer research.”
The caBIG™ Imaging Solutions: NCIA and caIMAGE
The NCIA and caIMAGE applications bring the goal of nationwide standardization of images closer to reality. caIMAGE is a database of cancer images that allows investigators and researchers to search for and submit images. The application utilizes the working vocabularies for clinical, translational, and basic research used in NCI data systems, thereby promoting standardization across institutions. caIMAGE is a web-based application that can be accessed through any browser after download, and requires minimal technical assistance for installation and use.
The NCIA is a searchable, national repository that integrates in vivo cancer diagnostic imaging studies with clinical and genomic data. It provides the cancer research community, industry, and academia with access to Digital Imaging and Communications in Medicine (DICOM) images, image markup, annotations, and rich meta data. Initial image libraries include the Reference Image Database to Evaluate Response (RIDER) and Lung Image Database Consortium (LIDC) datasets. The NCIA ultimately serves to increase the efficiency and reproducibility of imaging cancer detection and diagnosis, while enabling the development of imaging resources that will lead to improved clinical decision support.
“These tools and the efforts of the workspace may facilitate the building of consensus among many different vendors and academic institutions around how to share imaging information,” said Dr. Siegel. “Having a critical mass of these people all agreeing upon a way to do this and making it a free and open source is really extraordinarily powerful.”
With this standardization in place, physicians and researchers will be able to do many things, including rapidly search for cancer images with characteristics or features similar to those of a particular patient or study group, not just from their own databases, but from databases across the country. Access to this information could help them learn how previous patients responded to particular treatments and determine whether those same treatments might be applicable in other ways with current patients.
Sharing information in this way, for example, could help experimental drugs to move more rapidly toward approval or enable better prediction of which patients will respond best to which types of existing therapies.
“If we are successful, the use of tools such as NCIA and caIMAGE may benefit patients by allowing us as researchers and physicians to give them the right treatment for their specific case, rather than unnecessary trial and error to find cures,” said Dr. Siegel.
For more information about the caBIG™ In Vivo Imaging Workspace and its applications, visit: https://cabig.nci.nih.gov/workspaces/Imaging.
For the most current version of caIMAGE, visit: https://cabig.nci.nih.gov/tools/caIMAGE.
For the most current version of the National Cancer Imaging Archive (NCIA), visit: https://cabig.nci.nih.gov/tools/NCIA.
