An ontology-based framework to support nonintrusive storage and analysis of radiological diagnosis data

The purpose of the present work was to create a computational framework for radiological registry and diagnosis, which, by means of a nonintrusive approach, allows the users freedom annotation while storing related information in a structured and standard format. To get this goal, this work started with the investigation of lexical and semantic domain of radiological texts, followed by the design and implementation of an ontology, called RadOn, and the modeling of a database based on it. The next step was the development of a set of integrated software components, called E-Rad, based on text mining techniques, which identifies and extracts words, terms and expressions from free texts, adjusts them to the ontology structure and stores them. Finally, a query engine that works on the E-Rad structure, having a user interface, was implemented. The outcome is a computational environment that, by means of the implemented ontology and related tools, makes the data semantic explicit and allows the use of the context information in an optimal and standard way.