Efficient platform as a service (PaaS) model on public cloud for CBIR system

Abstract

Medical image processing requires handling a huge amount of data. Unstructured big data can create some issues related to latency. Distributed architectures based on parallelism can alleviate a latency problem. Also, achieving image availability in case of a server failure in such a huge system increases the latency time. To solve the challenges of latency in image processing in an enormous system, we propose a new platform for information retrieval in databases consisting of digital imaging communication in medicine (DICOM) files. The platform is based on a Decision Tree. The servers in this platform are distributed and work in a parallel way. Also, a fault tolerant system based on time triggered protocol is proposed to ensure image availability and minimize image recovery latency in the case of a server failure. The mam goal of this proposal is to select images from DICOM files similar to an image proposed in a query, using the principle of content based image retrieval (CBIR). Also, this platform helps radiologists with the diagnosis of medical images.