This article discusses the potential of Dicoogle Open-Source (www.dicoogle.com) for research and education in medical imaging informatics, focusing on using the Dicoogle Learning Pack framework. This tool provides documentation, configurations, and code examples to help users understand how Dicoogle resources teach the necessary understanding of medical imaging informatics using a modern server archive. This package also helps users navigate the system more efficiently.
Dicoogle has several key features that allow for new ways of looking into meta-imaging information for retrospective assessments. The system delegates all indexation tasks to indexer plugins, which are responsible for data indexing in a format that allows quick access to the stored information (Figure 1). These plugins also contain all the required procedures and dependencies for extracting and storing information from imaging files. This means that Dicoogle can provide data indexing and retrieval supported by non-relational databases, unlike many other medical imaging systems. In the research context, the project fosters high extensibility in its design and is a key component for many research and academic works.
Figure 1: Dicoogle General Architecture
Dicoogle was a pioneer in data mining over medical imaging repositories. It can identify inconsistencies in data and processes by enabling multiple views over the medical data repository and exporting data for further statistical analysis. This tool can be used to audit systems information data and contribute to the improvement of medical practice. The research developed in the last decade progressively pushes the use of Dicoogle towards the goal of multimodal medical information retrieval. Three main research topics were established, namely content-based image retrieval methods, a multimodal search engine for medical images, automated labeling systems, and representation learning methods.
In the academic context, Dicoogle has been used for fulfilling a variety of use cases in medical imaging informatics since its inception. In the last decade, twelve PhD and twenty Master students have used the Dicoogle framework in their works. At the University of Aveiro, a particular curricular unit is attended by radiology and engineering students. On a yearly basis, the department of Electronics, Telecommunications, and Informatics (DETI) offers Master’s and Doctoral degree students in computer science, as well as Master’s students in medical imaging technologies, the opportunity to enroll on the subject of Networks and Services in Imaging. The course teaches an average of 20 students every year. Prior to their enrollment, students of Medical Imaging Technology have limited knowledge of computer science. On the other hand, computer engineering students are not familiar with either the DICOM standard or the PACS concepts. The course teaches students about PACS and the DICOM standard. It addresses the main medical imaging modalities, quality management and control issues in PACS environments, and radiology information systems. Dicoogle enables those students and researchers to quickly prototype and deploy new functionality taking advantage of the embedded services. This full-fledged implementation of a medical imaging archive is very amenable to extension due to its plugin-based architecture and out-of-the-box functionality. This makes it very interesting for prototyping, experimentation, and bridging functionality with deployed applications.