Original Research Article Reproducibility of left and right ventricular mass measurements with cardiac CT Florian Schwarz, MD a,b , Richard Takx, BS a,c , U. Joseph Schoepf, MD a, * , Yeong Shyan Lee, MB, BCh, FRCR a,d , Balazs Ruzsics, MD, PhD a , Eva Maria Gassner, MD a,e , Salvatore Chiaramida, MD a , Thomas Henzler, MD a,f a Heart and Vascular Center, Medical University of South Carolina, 25 Courtenay Dr, Charleston, SC 29401, USA; b Institute of Clinical Radiology, Ludwig-Maximilians University, Campus Grosshadern, Munich, Germany; c Department of Radiology, Maastricht University Medical Center, Maastricht, The Netherlands; d Department of Diagnostic Radiology, Tan Tock Seng Hospital, Singapore; e Department of Radiology, University Hospital Innsbruck, Austria and f Institute of Clinical Radiology and Nuclear Medicine, University Medical Center Mannheim, Medical Faculty Mannheim, Heidelberg University, Germany KEYWORDS: Cardiac CT; Ventricular mass; Ventricular volumetric parameters; Semiautomatic software BACKGROUND: Cardiac CT provides volumetric data that enables characterization of the myocardium. OBJECTIVE: We evaluated intraobserver, interobserver, and interstudy reproducibility of left ventricular (LV) and right ventricular (RV) mass quantification with cardiac CT. METHODS: Thirty-eight patients who underwent cardiac CT twice within 365 days were included in this analysis. Functional reconstructions in 10% steps throughout the R-R interval and axial 1.5-mm sections were used. Semiautomatic contour detection was used to trace epicardial and endocardial borders in all cardiac phases for calculation of LV and RV ejection fraction, end-diastolic volume, end-systolic volume, cardiac output, stroke volume, and ventricular mass. For each study 2 observers measured LV and RV mass twice. RESULTS: LV mass parameters derived from semiautomatic contour detection algorithm had excel- lent intraobserver (r 5 1.00), interobserver (r 5 0.99), and interstudy (r 5 0.99) reproducibility (P , 0.0001). Average end-diastolic LV mass was 146.2 6 42.9 g at the first CT study and 146.8 6 44.4 g at the second study. For measuring RV mass, reproducibility was good on all levels (r 5 0.78, r 5 0.78, and r 5 0.68, respectively, with an average end-diastolic mass of 25.7 6 5.8 g at the first study and 24.4 6 4.8 g at the second study. CONCLUSION: Quantification of LV mass at cardiac CT with the threshold-based, region-growing semiautomatic segmentation analysis software evaluated here is highly observer independent and reproducible. This largely holds true for the estimation of RV mass as well; however, further improve- ments are needed to optimize reproducibility for RV mass quantification. Ó 2011 Society of Cardiovascular Computed Tomography. All rights reserved. Conflict of interest: U.J.S. is a medical consultant for Bayer-Schering, Medrad, and Siemens and receives research support from Bayer-Schering, Bracco, General Electric, Medrad, and Siemens. The remaining authors report no conflicts of interest. * Corresponding author. E-mail address: schoepf@musc.edu Submitted March 31, 2011. Accepted for publication August 3, 2011. 1934-5925/$ - see front matter Ó 2011 Society of Cardiovascular Computed Tomography. All rights reserved. doi:10.1016/j.jcct.2011.08.004 Journal of Cardiovascular Computed Tomography (2011) 5, 317–324