Reproducibility of tailored and universal nonselective excitation pulses at 7 T for human cardiac MRI : A 3‐year and an interday study

Abstract Purpose Ultrahigh‐field (UHF; ≥7 T) MRI is challenging due to spatially heterogeneous B 1 + profiles. This longitudinal study evaluates the reproducibility of three parallel‐transmission excitation strategies to enable UHF cardiac MRI: vendor‐supplied radiofrequency (RF) shim, subject‐tailored kT‐points pulses (TPs), and universal kT‐points pulses (UPs). Methods Six healthy subjects underwent 7 T MRI scans performed by different MR operators using a 32‐element parallel‐transmission body array at four time points over 3 years. A single UP was computed and applied to all subjects. TPs were computed individually for each scan and organized into four configurations. Each configuration was applied to all scans from each subject to analyze intrasubject variability. Reproducibility was assessed by comparing the coefficient of variation (CV) of simulated flip angles (FAs) within the heart volume across scan sessions. Results TPs designed for a specific scan session yielded lower CVs (2‐fold reduction) than UP. Applying TPs to other scan sessions of the same subject, however, resulted in approximately 40% higher CVs and lower FA uniformity compared with the UP. On average, the UP consistently achieved the most reproducible results across inter‐year, inter‐day, and same‐operator studies, with CVs of approximately 12%. Conclusion Although TPs showed advantages when tailored for a specific target volume, they struggled with long‐term consistency and required lengthy calibration. The precomputed UP kT‐points pulses proved to be the most consistent across all scans acquired in the 3 years by different operators, minimizing CV‐data dispersion and maintaining FA uniformity.