Adeel Siddiqui MBBS, Presenter: Nothing to Disclose

Ari Meir Blitz MD, Abstract Co-Author: Research Grant, B. Braun Melsungen AG Research Grant, Aesculap, Inc

In our study, we were able to create an SQL database of radiology reports after a one time data transfer. This created a research tool that could be used for multiple research projects. Investigators had a tailor made search engine that was suited for research purposes and more robust than traditional RIS/PACS search engines. Combing this approach with pathology and electronic health record information can lead to the formation of even better tools to identify patient cohorts and potential research targets.

Research work can be greatly increased if there are robust tools to search the radiology report database. This need is even more imperative if there is only a unique subset of patients or studies that research is being done for, and querying the entire PACS can become cumbersome. However, searching for specific reports within a RIS system is often met with limitations. These include but are not limited to lack of direct physician access to search reports, robustness of the RIS in handling system queries, and limitations in ways the information can be extracted from the RIS. The focus of this abstract was to see whether extraction of the radiology reports onto a separate SQL database would be an attractive alternative to query the pacs repeatedly for each research project.   

327 patients underwent a novel skull base MRI protocol between 2010 and 2012 at Johns Hopkins Hospital. The exact MRI protocol used are beyond the scope of this abstract. 515 MRI exams of 327 patients were successfully transferred from the RIS query to an SQL server running Microsoft SQL express. Two SQL tables were created. One table containing patient name, medical record number, etc and the second table containing the entire radiology report text linked to the Accession number. Using a dot-net and visual basic framework, a front end for entering patient data was created (fig 1).  

All the report text was easily searchable through the front end software. We were able to instantly identify separate cohorts such as patients with specific tumor types and also able to distinguish patients based on demographic factors such age, sex, and race. 

Siddiqui, A, Blitz, A, Using SQL to Create a Searchable Database of Radiology Reports for Research.  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13044268.html