Rowan Frank Walsh MBBCH, Abstract Co-Author: Nothing to Disclose

Yishay Salem MD, Abstract Co-Author: Nothing to Disclose

Wyman W Lai MD, Abstract Co-Author: Nothing to Disclose

James Nielsen MD, Presenter: Nothing to Disclose

The measurement error for right ventricular (RV) size and function assessed by cardiac MRI (CMR) in congenital heart disease (CHD) has not been fully characterized. Prior studies have utilized non-expert observers or subjects without significant disease involving the RV. Our goal was to examine the repeatability of CMR for RV size and function in subjects with CHD involving the RV.

Our CMR database was searched for 30 consecutive subjects with CHD involving the RV whose examination was performed by one of two CMR experts. CMR protocol for RV assessment was standard cine imaging in a ventricular short-axis plane with offline contouring of RV end-diastolic (RVEDV) and end-systolic volumes. Repeat contouring was performed by two expert observers to assess intraobserver (JCN, N=15) and interobserver (JCN versus WWL, N=15) error in two separate cohorts. The bias, coefficient of variability (COV) and repeatability coefficients (RC) were calculated. RC was multiplied by the mean value for each group studied to define a threshold beyond which measurement error was unlikely to be responsible.

Mean age = 18.9 +/- 10.7 yrs. Mean indexed RVEDV = 124 +/- 33 ml/m^2 (range 82 - 217 ml/m^2). Mean indexed RVEDV difference +/- SD (ml/m^2), COV (%) = 3.5 +/- 5.6, 3.1 and -4.3 +/- 5.4, 3.1 for intra- and interobserver comparisons, respectively. The repeatability coefficients were 12.7% and 13.3% for intra- and interobserver comparisons, which yielded threshold values of 14.3 ml/m^2 and 18.0 ml/m^2, respectively. For RV ejection fraction (EF), the RC for intra- and interobserver comparisons were 5.0% and 6.2%, which resulted in threshold values of 2.6 and 3.2 (EF%).

Expert CMR measurements of RV EDV and EF in patients with CHD have a reasonably high level of repeatability. For RVEDV, the threshold value--which defines the change between repeat measurements that is not due to measurement error with 95% certainty--is 14.3 ml/m^2 for intraoberserver comparisons and 18.0 ml/m^2 for interobserver comparisons.

The threshold values reported can be used when assessing serial differences in RV size and function. Low repeatability error confirms the utility of CMR in patients with CHD.

Walsh, R, Salem, Y, Lai, W, Nielsen, J, Repeatability of Cardiac MRI Measured Right Ventricular Size and Function in Congenital Heart Disease.  Radiological Society of North America 2009 Scientific Assembly and Annual Meeting, November 29 - December 4, 2009 ,Chicago IL. http://archive.rsna.org/2009/8008835.html