Fast, precise, and accurate myocardial T1 mapping using a radial MOLLI sequence with FLASH readout
During the last decade, the improvement of hardware systems, along with the development of faster acquisition schemes, have enabled quantitative measurements of myocardium T1 and T2 relaxation times 14 and their use in the monitoring of disease progression or therapy response. Quantitative CMR is even becoming a standard imaging tool in clinical trials 15. Whereas qualitative imaging is routinely used to detect focal and intense lesions through late‐gadolinium enhancement 16, quantitative T1 mapping sequences could reveal diffuse increases of gadolinium (Gd) chelates volumes of distribution, such as in patients suffering from cardiomyopathies 17 and amyloidosis 18. Subtle variations of native T1 relaxation times were also quantified in acute myocardial infarction 19 or Anderson‐Fabry disease 20.
The first clinically acceptable method that was developed to perform myocardial T1 mapping was introduced more than 10 years ago and is called the modified Look‐Locker inversion recovery (MOLLI) sequence 21. This pioneer sequence has been widely applied because of a high degree of reproducibility and precision, but it has also been shown to be slightly dependent on heart‐rate variations 22. Its acquisition scheme was then optimized and the current 533 version appears to be the most popular in overcoming this limitation. Other methods were more recently developed like shMOLLI 23, SASHA 24, or SAPPHIRE 25, resulting in T1 maps with better accuracy but poorer precision than MOLLI on T1 estimates 26. This effect could be attributed for some sequences to a smaller number of data points sampled during the T1 relaxation process, or to the application of a saturation instead of an inversion pulse for T1 weighting. Indeed, all of the aforementioned strategies are based on the acquisition of several images (8 to 11) during a single breath‐hold at different time points after application of a T1‐weighted preparation pulse. The T1 value is computed from these images by adjusting the magnetization recovery with exponential models. Depending on the method, the acquisition time of a single slice varies between 9 and 17 heartbeats. Moreover, it has been reported that long breath‐holds are associated with a linear drift in the foot–head direction in the order of 0.4 mm/s, causing possible misregistration of the different images while computing the T1 maps 27. This limitation, besides the fact that long breath‐holds are generally not well tolerated by patients, led to the development of novel approaches based on free‐breathing acquisitions. Weingartner et al, for instance, introduced a slice‐interleaved sequence called STONE, allowing the quantification of native T1 relaxation times and resulting in a scan time of less than 20 s per slice 28.
Most of the T1‐mapping sequences rely on single‐shot true fast imaging with steady‐state‐precession (TrueFISP) readouts that allow producing images with higher signal‐to‐noise ratio (SNR) as compared with fast low‐angle shot (FLASH). Nevertheless, TrueFISP is known to be more sensitive than spoiled gradient‐echo readouts to susceptibility, B0 and B1 field inhomogeneities, as well as T2 relaxation variations 29. These parameters are not taken into account when using exponential models to determine T1, and led to biased measures. For example, the MOLLI sequence is known to underestimate high T1 values (with more than 15% underestimation for T1 values higher than 1500 ms).