Description:
Summary
Harnessing the power of large databases is contingent upon the ability to access datapoints efficiently and accurately; however, this has proven difficult in the field of radiology where complex visual and written records must be integrated. Oregon Health & Science University research has led to the development of a novel radiology-specific search engine and report classification method, RadSearch, to allow for quick and accurate queries of radiology reports providing the potential for improved patient care, education and research.
Technology Overview
Radiology is well-suited for artificial intelligence (AI) assisted analysis, given these health records contain complex and paired linguistic and visual data. However, previous attempts to unlock this information have been encumbered by heterogeneity of human language, proprietary search algorithms, and lack of medicine-specific search performance matrices. OHSU researcher Dr. Peter Li has developed a radiology-specific, secure, and efficient search engine, RadSearch, to allow doctors to quickly and efficiently query reports. Utilization of this tool would allow radiologists to quickly and efficiently search for relevant positive or negative reports, thereby tapping into the full-power of their healthcare database.
Features of the RadSearch tool include:
- Standalone search engine independent of specific PACS system;
- Fast and accurate intuitive keyword query of radiology reports;
- Built-in natural language processing for classifying queried reports as positive, negative, or neutral for specific radiological findings with 95% accuracy during initial testing;
- Standalone credential verification for increased security and minimized risk of data breach.
Ultimately this tool will provide better access to radiology data and has the opportunity to refine educational content, improve patient care, advance clinical research, and streamline quality assurance and performance monitoring.
Publication
Li, N., Maresh, G., Cretcher, M., Farsad, K., Al-Hakim, R., Kaufman, J., & Gichoya, J. (2020). A modern non-SQL approach to radiology-centric search engine design with clinical validation. arXiv preprint arXiv:2007.02124. Link
Licensing Opportunity
This technology is available for licensing.