1357700 - Assessment of ventricular function using a new retrospective k-t cine acceleration sequence

An accurate and reproducible retrospective 2D cine balanced steady-state free precession (2D-SSFP) sequence is the preferred method in clinical practice, the gold standard modality for non-invasive quantitative evaluation of left ventricle (LV) dimensions, function and myocardial mass. But the long acquisition and breathholds may not be possible for critically ill patients.
To avoid slice misregistration and image artifacts that may preclude adequate volumetric estimation, accelerating cine acquisition without compromising spatial or temporal resolutions has received constant research interest. At the beggining, a prospective gating was used with this accelerating technique. But in this study, our goal was to validate a recent retrospective k-t 2D SENSE-based accelerated cardiovascular magnetic resonance (CMR) sequence (Retro-KT) in symptomatic heart failure patients, assessing this technique´s accuracy for quantifying left ventricle (LV) volumes, mass, function and global circumferential strain (GCS) as compared to standard 2D cine steady-state free precession (2D-SSFP).

Methods:

Volunteers with no prior cardiovascular symptoms (n=10) and consecutive symptomatic heart failure patients (n=27) were scanned using a 1.5T MRI system for LV function assessment. Two LV Short axis stacks (SA) were acquired: Retro-KT (5-6 breath-holds; temporal/spatial resolution: 19.2ms/1.0x1.0mm; acceleration factor of 8) and standard 2D-SSFP (10-12 breath-holds; temporal/spatial resolution: 27.2ms/2.0x2.0mm, parallel imaging). LV stroke volume (LVSV) was assessed on all volunteers through the ascending aorta phase-contrast evaluation, for the Retro-KT sequence validation. An image quality score for both stacks was applied, with artifact burden quantification.

Results:

Patients had lower mean LV ejection fraction (LVEF, 41±18%, p< 0.001) and higher LV indexed volumes and mass than volunteers (p-values< 0.05). There was a strong correlation between sequences for LV measurements, with high agreement: mean differences or bias (limits of agreement) were 1.2% (-6.68% to 9.08%), -0.05 mL/m² (-10.78 to 10.68 mL/m²), -1.49 (-11.68 to 8.7 mL/m²), 0.34 g/m² (-12.75 to 13.43 g/m²) and -0.92% (-7.39 to 5.55%) for LVEF, indexed end diastolic/systolic volumes and mass, and GCS respectively. LVSV by Retro-KT presented a strong correlation and good agreement with aortic flow. SA acquisition duration was reduced by nearly 50% by the Retro-KT cine sequence compared to 2D-SSFP (74±17s versus 146±25s, p< 0.001).

Conclusion:

Retro-KT cine sequence is feasible and reproducible, allows accurate LV volumes, mass and myocardial strain assessment with high diagnostic image quality, and considerably shortening acquisition time, even in critically ill patients.