Benoit Desjardins1, R Curtis Hamilton. 1. Department of Radiology, University of Michigan, Ann Arbor, MI 48109-0326, USA. benoitd@umich.edu
Abstract
RATIONALE AND OBJECTIVES: We present a method to perform full text searches of radiology reports for the large number of departments that do not have this ability as part of their radiology or hospital information system. MATERIALS AND METHODS: A tool written in Microsoft Access (front-end) has been designed to search a server (back-end) containing the indexed backup weekly copy of the full relational database extracted from a radiology information system (RIS). This front end-/back-end approach has been implemented in a large academic radiology department, and is used for teaching, research and administrative purposes. RESULTS: The weekly second backup of the 80 GB, 4 million record RIS database takes 2 hours. Further indexing of the exported radiology reports takes 6 hours. Individual searches of the indexed database typically take less than 1 minute on the indexed database and 30-60 minutes on the nonindexed database. Guidelines to properly address privacy and institutional review board issues are closely followed by all users. CONCLUSIONS: This method has potential to improve teaching, research, and administrative programs within radiology departments that cannot afford more expensive technology.
RATIONALE AND OBJECTIVES: We present a method to perform full text searches of radiology reports for the large number of departments that do not have this ability as part of their radiology or hospital information system. MATERIALS AND METHODS: A tool written in Microsoft Access (front-end) has been designed to search a server (back-end) containing the indexed backup weekly copy of the full relational database extracted from a radiology information system (RIS). This front end-/back-end approach has been implemented in a large academic radiology department, and is used for teaching, research and administrative purposes. RESULTS: The weekly second backup of the 80 GB, 4 million record RIS database takes 2 hours. Further indexing of the exported radiology reports takes 6 hours. Individual searches of the indexed database typically take less than 1 minute on the indexed database and 30-60 minutes on the nonindexed database. Guidelines to properly address privacy and institutional review board issues are closely followed by all users. CONCLUSIONS: This method has potential to improve teaching, research, and administrative programs within radiology departments that cannot afford more expensive technology.