Medical image archives — present and future

Abstract

Repositories, data warehouses, public archives, and multimedia databases, especially those containing images, are important for clinical practice, education, and research in medicine. Imaging modalities, such as digital radiography, computed tomography, ultrasonography, nuclear medicine, digital angiography and magnetic resonance imaging produce large amounts of data that are managed, archived, and reported using picture archiving and communications systems (PACS) in an integrated healthcare enterprise (IHE) infrastructure. The technology and applications of PACS and IHE are well understood and widely available worldwide.

However, typical PACS and IHE systems are closed to ensure privacy, security and confidentiality. Sharing of medical image databases is relatively new, despite the widespread prevalence of open public archives for other types of biological data, especially in genomics and proteomics. Bioinformatics often implies that sequence and not image information is the focus of interest. Medical image visualization, analysis and measurement are widely used, and should continue to grow in importance as computer and display technology continues to evolve.

The principles of medical image archives, repositories, data warehouses, and multimedia databases will be reviewed. The applications of publicly available data sets include research, both basic and clinically applied, on the development, testing and standardization of algorithms and heuristics for computer assisted diagnosis, for example. Data mining of clinical databases, especially those richly invested with images, may be beneficial in producing quantitative image-based metrics and biomarkers for disease presence, staging, treatment selection and planning, prognosis, and monitoring. The technical requirements for construction of publicly accessible medical image archives that are well suited to exploratory data analysis, synthesis and testing of diagnostic and prognostic quantitative image analysis methods, biomarker and surrogate endpoint testing, and computer aided diagnosis will be outlined.

Regulatory requirements for testing of quantitative medical image-based analysis methods are in development. The FDA requirements for electronic submission and review of medical image analysis devices will be described, and their impact on the technical requirements for publicly accessible medical image archives will be delineated.