Researchers at MD Anderson Cancer Center in Houston have developed ViSion, a multimedia structured reporting software application that can automate the process of calculating the quality of radiology reports.
ViSion captures key images and a radiologist's verbal descriptions of image findings, according to presenter Dr. David Vining. It also tags the images -- using natural language processing -- with metadata that are referenced to an ontology (i.e., a controlled vocabulary with defined relationships between terms) and then assembles a multimedia report with image findings linked to historical data in timelines. Each ontology term is tagged with "priority" and "specificity" values.
"By analyzing the values of connected data in the timelines, we can generate a quality score for a radiology report," he said.
The researchers believe their reporting system can be used to measure the quality of radiology reports and, more importantly, improve their information content prior to completion.
"ViSion incorporates 'common data elements' in each finding (i.e., discrete elements that are required to describe a disease) and conducts a compliance check at the time of report signoff to remind a radiologist to complete those details if necessary," Vining told AuntMinnie.com. "The goal is to help radiologists produce more-complete and higher-quality radiology reports, and this, in turn, should benefit radiologists as we move toward a pay-for-performance reimbursement environment."
Visit the Lakeside Center to delve further into this poster presentation.
