A distributed architecture for content-based image retrieval in medical applications
Mark Oliver Güld, BERTHOLD B. WEIN, Daniel Keysers, Christian J. Thies, Michael Kohnen, Henning Schubert, Thomas Lehmann · 2002
Image retrieval in medical applications (IRMA) incorporates knowledge from the fields of medicine, image analysis for diagnostic purposes and system engineering. Its implementation as a distributed development platform is fundamental for an efficient interdisciplinary knowledge transfer. The distributed IRMA architecture provides location and access transparency for its resources, i.e. images, feature vectors and methods, resulting in automatic distribution to all participating work groups, including automated replication functionality. The necessary administration is done via a central database with special attention to automated replication functionality. Concurrency transparency and automatic distribution of tasks for image processing, feature extraction, feature evaluation and classification allow the utilization of the computational power of all IRMA integrated hosts regardless of their operating system or hardware configuration. Via extensive system transparency, IRMA drastically simplifies the cooperation of the interdisciplinary development team, allowing all partners to focus on their expert field. In particular, this vastly improves communication and evaluation processes, resulting in much shorter development cycles for new medico-diagnostic methods.