A GENERAL NATURAL-LANGUAGE TEXT PROCESSOR FOR CLINICAL RADIOLOGY

Objective: Development of a general natural-language processor that id entifies clinical information in narrative reports and maps that infor mation into a structured representation containing clinical terms. Des ign: The natural-language processor provides three phases of processin g, all. of which are driven by different knowledge sources. The first phase performs the parsing. It identifies the structure of the text th rough use of a grammar that defines semantic patterns and a target for m. The second phase, regularization, standardizes the terms in the ini tial target structure via a compositional mapping of multi-word phrase s. The third phase, encoding, maps the terms to a controlled vocabular y. Radiology is the test domain for the processor and the target struc ture is a formal model for representing clinical information in that d omain. Measurements: The impression sections of 230 radiology reports were encoded by the processor. Results of an automated query of the re sultant database for the occurrences of four diseases were compared wi th the analysis of a panel of three physicians to determine recall and precision. Results: Without training specific to the four diseases, r ecall and precision of the system (combined effect of the processor an d query generator) were 70% and 87%. Training of the query component i ncreased recall to 85% without changing precision.