Abstract 12255: Prognostic Signals From Moderate Aortic Valve Disease: An Artefact of DICOM Structured Reporting?

Abstract

Background: The use of Big Data in cardiac imaging research is usually based on DICOM structured reporting (SR) files. Default export settings from SR include all measurements, which may be misleading when diagnosis is based on peak measures. Validation of research datasets and remeasurements are seldom performed. We hypothesized that use of DICOM-SR outputs may lead to misclassifying patients with severe aortic stenosis (AS) as moderate. Methods: AS cases were filtered from an echocardiography dataset based on aortic valve area (AVA) <2 cm 2 , and peak velocity (V max ) >2 m/s. Data validation was conducted i) by verifying each parameter, ensuring only appropriate values were considered and peak gradients acquired, and compared with ii) remeasurement. Outlier and erroneous captured values were excluded. Values of each parameter and definitions of AS were compared pre- and post-validation and remeasurements. Results: Of 919 validated echocardiograms with moderate or severe AS, up to 60% of echocardiographic parameter values, such as AVA, mean and peak gradient, or dimensionless severity index (DSI) had minimum 10% error pre- and post-validation (Table). Based on 2 of 5 criteria (AVA, AVAi, V max , MPG, and DSI) for classifying AS severity, 42% of echocardiograms changed degree of AS severity. Among echos identified as moderate AS pre-validation, 173/413 (41%) were reclassified as severe AS. The main reason for erroneous data was averaging of all repeated echo measures. These differences exceeded changes due to remeasurement in a subgroup of 262 echocardiograms by an order of magnitude (Table). Conclusions: Non-peak data in DICOM-SR led to incorrect categorisation of a high proportion of patients based on AS severity definitions. Standardisation of data fields, and curation to ensure that only peak values are considered in DICOM-SR data are essential. Table . Differences in echocardiographic parameters pre- and post-validation of DICOM-SR or remeasurement