ABSTRACT

In this paper, we propose a text mining system to extract and use the information in radiology reports. The system consists of three main modules: medical finding extractor, report and image retriever. The medical finding extraction module automatically extracts medical findings and associated modifiers to structure radiology reports. The structuring of the free text reports bridges the gap between users and report database, makes the information contained in the reports readily accessible. It also serves as intermediate result to other components of the system. The retrieval module analyzes user’s query and returns the reports and images that match the query. The overall evaluation results are satisfactory, though more thorough testing and evaluation are needed. Our future work includes improving the current system performance and implementing the radiology report generation system using statistical machine translation approach, for which we have designed the general architecture..