Knowledge-based compilation of magnetic resonance diagnosis reports in neuroradiology

The compilation of the neuroradiology diagnosis reports based on magnetic resonance (MR) exams comprises a deep analysis of images and related numerical values, usually done with specialized image processing tools, and the compilation of the different parts forming the reports, following well-defined schemes which depend on the kind of exam and pathology. Although the diagnosis report is a semi-structured document comprising different well-defined parts, usually it is compiled using simple text editors that may loose its structure. The drawback of this approach is twofold, first of all the specialist has to repeat the writing of some texts for each report, yielding to a time consuming process, and second and most importantly, the precious data, annotations and comments written in the report are not easily available for further analysis. In fact, when the information contained in the diagnosis reports is stored into unstructured documents such as texts, it is very difficult to query and extract useful information needed for conducting studies on large populations of patients. In this paper we propose a novel software tool able: (i) to store neuroradiology diagnosis reports and their schemes in a structured knowledge-base; and (ii) to support the specialist in the compilation of new diagnosis reports on the basis of the schemes and contents already stored in the knowledge-base.