NoSQL distributed database for DICOM objects

Every year, the amount of data produced in health-care institutions in the field of medical imaging is increasing. The easy accessibility to digital modalities and the proliferation of diagnosis centers working in network of services, creates the need for very large repositories of data. Moreover, the institutions have redundant storage across the network and load balancing mechanism to improve access performance. NoSQL databases provide horizontal scalability which allows the system to grow easily and effortlessly. Document databases are particularly suitable to store the DICOM metadata and the JSON format is already supported by the standard. This paper, we discuss the use of document database model in the medical imaging field, trying to focus on both clinical and research environments. Moreover, we focused on the proposal of an architecture based on MongoDB to distribute the load between multiple nodes. The proposal was integrated with an open-source PACS and validated by simulating multiple operations over data distributed across multiple virtual locations, and the results are promising.