@Article{chd.12484,
AUTHOR = {Elise D. Pieterman, Ricardo P. J. Budde, Danielle Robbers-Visser, Ron T. van Domburg, Willem A. Helbing},
TITLE = {Knowledge-based reconstruction for measurement of right ventricular volumes on cardiovascular magnetic resonance images in a mixed population},
JOURNAL = {Structural and Congenital Heart Disease},
VOLUME = {12},
YEAR = {2017},
NUMBER = {5},
PAGES = {561--569},
URL = {http://www.techscience.com/schd/v12n5/39189},
ISSN = {3071-1738},
ABSTRACT = {<b>Objective:</b> Follow-up of right ventricular performance is important for patients with congenital
heart disease. Cardiac magnetic resonance imaging is optimal for this purpose. However, observerdependency of manual analysis of right ventricular volumes limit its use. Knowledge-based reconstruction is a new semiautomatic analysis tool that uses a database including knowledge of right
ventricular shape in various congenital heart diseases. We evaluated whether knowledge-based
reconstruction is a good alternative for conventional analysis.<br/>
<b>Design:</b> To assess the inter- and intra-observer variability and agreement of knowledge-based versus conventional analysis of magnetic resonance right ventricular volumes, analysis was done by
two observers in a mixed group of 22 patients with congenital heart disease affecting right ventricular loading conditions (dextro-transposition of the great arteries and right ventricle to
pulmonary artery conduit) and a group of 17 healthy children. We used Bland-Altman analysis and
coefficient of variation.<br/>
<b>Results:</b> Comparison between the conventional method and the knowledge-based method
showed a systematically higher volume for the latter group. We found an overestimation for enddiastolic volume (bias -40 ± 24 mL, r = .956), end-systolic volume (bias -34 ± 24 mL, r = .943),
stroke volume (bias -6 ± 17 mL, r = .735) and an underestimation of ejection fraction (bias 7 ±
7%, r = .671) by knowledge-based reconstruction. The intra-observer variability of knowledgebased reconstruction varied with a coefficient of variation of 9% for end-diastolic volume and
22% for stroke volume. The same trend was noted for inter-observer variability.<br/>
<b>Conclusion:</b> A systematic difference (overestimation) was noted for right ventricular size as
assessed with knowledge-based reconstruction compared with conventional methods for analysis.
Observer variability for the new method was comparable to what has been reported for the right
ventricle in children and congenital heart disease with conventional analysis.},
DOI = {10.1111/chd.12484}
}