Reproducibility of T1 mapping 11-heart beat MOLLI Sequence
Journal of Cardiovascular Magnetic Resonance volume 17, Article number: W26 (2015)
T1 mapping measuring the longitudinal relaxation time constant is increasingly used clinically for quantification of interstitial myocardial fibrosis, with higher myocardial T1 and ECV associated with adverse prognosis (Wong 2012). A working group consensus statement (Moon 2013) has recommended amongst other:
Use of native T1 myocardium and ECV only, avoiding partition coefficient (λ)
Use of gadolinium-based contrast agents (GBCA) for ECV calculation routinely and post-gad T1 maps taken at least 15 minutes following administration
Use of at least 2 different slice orientations to add diagnostic confidence
We sought to investigate the reproducibility of an 11-cycle 5(3)3 MOLLI sequence (Siemens prototype WIP448B) and furthermore wished to investigate whether repeating the MOLLI acquisition on one slice orientation twice could increase precision.
15 healthy volunteers (31±5 years, 8 males) with no known medical conditions, on no medication, underwent CMR scans (Avanto, Siemens, 1.5T) on two separate attendances with an 11 heart beat MOLLI 5(3)3. Following frequency adjustment, native T1 maps were obtained twice on a basal and a mid-ventricular slice respectively. 15 mins following GBCA administration (0.1 mmol/kg, Gadobutrol, Bayer, Germany) both image planes were acquired again twice. Pixel-wise T1 maps were analyzed offline by 2 independent blinded operators. A Region Of Interest was manually drawn in the septum for myocardium and blood in native and post gad images as shown in fig 1.
Intrastudy and interstudy intraclass coefficient (ICC) was assessed for various models (model A-F) combining an increasing number of basal and mid level acquisitions as shown in table 1. Intrastudy ICC was excellent for native T1 myocardium and ECV (0.94 and 0.86), but not for λ (0.54). Interstudy ICC for native T1 was very good for all the models [highest model E (0.90), lowest model B (0.77)]. Interstudy ICC for λ was overall poor [highest model F (0.79), lowest model B (0.70)]. For ECV there was excellent interstudy ICC when at least 2 native and 2 post-gad slices were included, best model F ( 0.93; with 1 basal and 1 mid level acquisition repeated, total 4 acquisitions native and 4 post gad). Interobserver reproducibility was excellent with intra and inter scan reproducibility across all models for T1, λ and ECV >0.90 (highest model F ECV ICC 1.00, lowest model E λ ICC 0.94).
Interpreted together these results suggest an 11 heart beat MOLLI prototype sequence can have high intra and inter scan and interobserver reproducibility. ECV shows the best reproducibility especially when images of one basal and one mid level are repeated twice (model F) and that T1 native myocardium reproducibility is excellent independently of model used. We would therefore recommend that if only 2 levels are acquired with the MOLLI sequence a repeat acquisition should be considered at each slice, to increase precision and reproducibility.
NIHR BRU, Royal Brompton Hospital, London.
About this article
Cite this article
Vassiliou, V., Heng, E.L., Sharma, P. et al. Reproducibility of T1 mapping 11-heart beat MOLLI Sequence. J Cardiovasc Magn Reson 17 (Suppl 1), W26 (2015). https://doi.org/10.1186/1532-429X-17-S1-W26