Clinical application of free-breathing compressed sensing in cardiac magnetic resonance cine sequences for assessing left ventricular function and strain.

Background: Cardiac magnetic resonance (CMR) is essential for evaluating cardiac function but is limited by its long scan times and the requirement for breath-holding. This has prompted the development of accelerated techniques such as compressed sensing (CS); however, most studies on CS have been conducted under breath-hold (BH) conditions. Thus, this study aimed to investigate the feasibility of CMR cine with free-breathing (FB) CS by comparing the left ventricular (LV) function and strain parameters of this technique with those obtained using conventional BH cine. Methods: Thirty-four patients who underwent CMR examination at Tongji Hospital were prospectively enrolled in this study from January 2023 to August 2023. All patients underwent cardiac short-axis and four-chamber long-axis scans with conventional BH cine and FB CS cine sequences. Two radiologists evaluated the overall image quality of the two cine sequences via a 5-point scale. The parameters of LV function and strain were calculated. Paired t-tests or Wilcoxon signed-rank tests were used to assess differences in quantitative data and qualitative scoring data between the two sequences. Intraclass correlation coefficient (ICC) and Bland-Altman analyses were performed to assess the agreement of the quantitative parameters obtained from the two sequences. Results: The duration of FB CS cine scans (14.8±2.0 s) was significantly shorter than that of conventional BH cine scans (83.6±10.9 s; P<0.001). The overall image quality of FB CS cine (4.0±0.6) was comparable to that of conventional BH cine (4.1±0.7) (P=0.561). There were no significant differences in the LV end-diastolic volume (EDV), cardiac output (CO), end-diastolic mass (EDM), or time to peak longitudinal strain (TPLS) between the two sequences (all P values >0.05), whereas the LV end-systolic volume (ESV), stroke volume (SV), ejection fraction (EF), global peak radial strain (GRS), global peak circumferential strain (GCS), global peak longitudinal strain (GLS), time to peak radial strain (TPRS), and time to peak circumferential strain (TPCS) were significantly different (all P values <0.05). There was good agreement in all the function and strain parameters between the two sequences according to the ICC analysis (all ICCs >0.84; all P values <0.001). Bland-Altman analysis revealed strong agreement in all the quantitative parameters between the two cine sequences, with a mean difference close to zero and a small range of variation. Conclusions: Compared with the conventional BH cine method, the FB CS cine method achieved comparable image quality with shortened scan times and good agreement in terms of LV function and strain parameters, supporting its clinical application.