Measurements and Normative Datasets in Congenital Cardiac MRI: A Survey by the Pediatric and Congenital Heart Disease Section QI Subcommittee

Uniform quantification of CMR data in pediatric and congenital heart disease (PCHD) would improve diagnostic and prognostic consistency, yet multiple measurement techniques and normative datasets are used. By identifying inconsistencies, targeted improvement efforts can be made.

Methods:

The PCHD Quality Improvement Subcommittee developed and vetted a REDCap survey, and distributed it via the SCMR listserv and WhatsApp group (April–May 2025). Percentages >100% reflect multiple selectable responses.

Results:

Responses from 44 centers in the United States, Europe, South America, and Asia were received. The most common annual CMR volume was 250–500 (37%). CMR interpretation was mainly by cardiologists (65%), with combined cardiology–radiology reading in 21%.

All centers use short-axis imaging for left ventricular (LV) measurements; 80% include papillary muscles in the blood pool. Most (84%) use short-axis for right ventricular (RV) volumes, with axial and 4-chamber views also used. Half do not routinely measure RV mass. Among those who do, 42% exclude trabeculations/bands and 26% include only major ones connected to a wall (Fig. 1)

Most centers (77%) report ventricular size/mass relative to normative data. The pooled datasets of Kawel-Boehm et al were most common (64%), followed by Buechel et al (41%) and Olivieri et al (27%). Nine other datasets were also used, including non-CMR based. Eighty-seven percent use different datasets for adults vs. children. Ventricular data are reported using Z-scores (57%) and/or normal ranges (57%).

Aortic measurements are most often made from non-contrast 3D SSFP (77%) and contrast MRA (61%). Other sequences include cine short-axis, cine long-axis, and black-blood (Fig. 2A). Measurements are performed in diastole (52%), systole (43%), or ungated MRA (25%), with double-oblique cross-sectional views used by 89%.

Seven methods are used for aortic root assessment (Fig. 2B). The most common are sinus-to-sinus on 3D datasets (48%) and sinus-to-commissure on 3D datasets (43%). Additional methods include cine short-axis in systole or diastole, and LV outflow long-axis views.

Non-CMR normative datasets (echo or CT) are used most often for aortic data (39%). Kawel-Boehm datasets are used by 34%, Kaiser et al by 23%, and Voges et al by 11%. Most centers use different pediatric vs. adult datasets (68%) and report aortic data using Z-scores (64%) and/or normal ranges (30%).
Other structures assessed with normative data include pulmonary arteries (73%), right atrium (30%), and left atrium (39%). Most centers manually generate Z-scores (57%). Z-scores are the most common severity-grading system (64%), with typical thresholds of mild 2–4, moderate 4–6, and severe >6 (56%).

Conclusion:

Significant variability exists in CMR quantification practices in PCHD, especially in RV mass and aortic measurements, with a wide range of datasets used, including non-CMR based. These findings will help guide efforts in standardization of measurement techniques and quantification.

Relative Importance of Potential Improvement Efforts, as Reported by Responding Centers