Algorithm automatically extracts breast density information from mammo reports

Tuesday, November 29 | 12:45 p.m.-1:15 p.m. | LL-INS-TU4B | Lakeside Learning Center

A method of automatically extracting breast density from free-text mammography reports is described in this poster presentation. Over time, this data could facilitate epidemiological research and decision support for patient management based on breast density.

Dr. Daniel Rubin, an assistant professor of radiology and medicine at Stanford University School of Medicine, will explain how an algorithm to automatically detect and extract BI-RADS breast density classes works.

The algorithm classifies each report into one of four density classes: fatty, fibroglandular, heterogeneously dense, or extremely dense. Developed by Rubin and colleagues and by researchers from the department of radiology at the University of Wisconsin, the algorithm uses pattern matching and regularly used phrases in unstructured commentary in mammography reports.

The algorithm was tested with 300 mammography reports from each institution; it achieved 99% and 100% accuracy. Additional testing is planned.