Assessing cardiac 2D flow MRI reliability across various physiologic, technical, and observer confounders in healthy traveling volunteers

PURPOSE: To evaluate the repeatability and reproducibility of phase-contrast two-dimensional blood flow cardiac MRI (2D flow) measurements in the presence of physiologic, technical, and observer confounders. MATERIALS AND METHODS: In this prospective observational study from January to June 2022, 20 healthy volunteers underwent 2D flow cardiac MRI at four sites (one 1.5-T and three 3-T scanners). Each participant received two consecutive segmented gradient-echo acquisitions and one real-time sequence with a scan-rescan examination at one site. Intraclass correlation coefficients (ICC) assessed sequence repeatability, scan-rescan reproducibility, sequence-type comparison, field-strength and scanner-configuration effects, and interobserver agreement. Correlations between forward flow volume (FF), peak velocity (PV), heart rate, and blood pressure were calculated using Pearson and Spearman coefficients. Segmented 2D flow cardiac MRI was compared with four-dimensional flow cardiac MRI at the sinotubular junction and aortic valve. RESULTS: Nineteen participants (mean age, 26 years ± 6 [SD]; 11 male participants) were analyzed. FF and PV demonstrated excellent sequence repeatability (ICC, 0.986 and 0.969, respectively) and scan-rescan reproducibility (ICC, 0.976 and 0.903, respectively). Different sequence types and field strengths showed excellent agreement for FF (ICC for sequence type, 0.925; ICC for field strength, 0.918), as did different scanner configurations and interobserver analyses with measurements within equivalence limits, except for PV across sequence types, field strengths, and between sites 2 and 3. Heart rate and FF correlated mildly (r, −0.40) but not systolic or diastolic blood pressure with FF and PV and heart rate with PV (r, −0.02). 2D and four-dimensional flow cardiac MRI correlated and had good to excellent agreement with FF at the sinotubular junction and aortic valve (r, 0.86; ICC, 0.930 and r, 0.89; ICC, 0.939, respectively) and moderate for PV (r, 0.77; ICC, 0.862 and r, 0.66; ICC, 0.767, respectively). CONCLUSION: 2D flow cardiac MRI measurements showed excellent repeatability for FF and moderate agreement for PV across sequences, field strengths, and observers. Physiologic and technical confounders had minimal impact on measurement precision.