Development of a Search Engine for Radiology Reports

Background and Aim: Radiological reporting has generated large quantities of digital content within the electronic health record. This information has potential for improving clinical care and supporting research. Easy-to-use tools are required to benefit from this information. We have developed a user-friendly search engine for radiology reports.

Methods: The Sheba Medical Center Big-data and artificial intelligence (AI) ARC team has developed the search engine. Reports were retrieved from the Sheba Medical Center radiology information system (RIS). A backend dedicated Structured Query Language (SQL) database was used to store and index the reports for fast retrieval. The frontend user interface was written in Java. Search results are presented in the user interface and can be downloaded in Microsoft Excel format.

Results: Overall, more than 2 million radiology reports were indexed. Data included radiology and nuclear medicine reports. Retrieval time was less than a second for queries of up to 100,000 reports. Data could be retrieved in anonymized or non-anonymized form. Search query included an advanced option for binary logic. Additional search filters include demographics, ICD-10 coding, radiology modality type and body part, signing radiologist, and ordering unit.

Conclusion: Easy-to-use and straightforward radiology reports search engine has been developed. This system can help create large research cohorts, identify important clinical cases, and aggregate operational data.