Synopsis

Magnetic Resonance Fingerprinting is an emerging technique for the simultaneous collection of T1 and T2 maps. Cardiac MRF is also possible, and can be used to make multiparametric maps of one or more 2D slices of the heart in 5 to 10 heartbeats.

Magnetic Resonance Fingerprinting (MRF) has recently emerged as a technique that can simultaneously map T₁, T₂, and proton density values in a single short (<10s) scan [1]. In MRF, the pulse sequence parameters (typically the flip angles and T_Rs) are varied during the scan, which prevents the magnetization from reaching a steady state. Tissues with different underlying T₁ and T₂ values will exhibit signal evolutions that are different from one another. These signal evolutions are complex and difficult to fit to a simple signal model, such as an exponential recovery or decay, as would be done in conventional parameter mapping. Instead, the experimental signal evolution in each pixel is matched to a dictionary of possible signal evolutions calculated using the Bloch equations, knowledge of the pulse sequence, and an input range of T₁ and T₂ values. While it would be possible to perform tissue property mapping in this way using fully-sampled data, the acquisition time would be long (approximately several minutes for a 2D scan). In MRF, a highly undersampled spiral trajectory is employed to dramatically reduce the scan time. As long as the aliasing artifacts produced by the undersampling are incoherent, pattern matching can still be used to extract the underlying T₁ and T₂ values despite the additional interference.

MRF has been used to generate T₁ and T₂ maps in the brain [1,2], prostate [3], and body [4] in both 2D and 3D. While first used to map relaxation values, MRF can be employed to map any tissue property to which a sequence is sensitive, diffusion [5], perfusion [6,7], partial volume effects [8], and microvascular parameters such as blood oxygenation [9] . However, MRF relies on the idea that the underlying tissue properties do not change during the acquisition, and thus the tissues cannot be in continual motion during the scan. While the reconstruction is less sensitive to motion that occurs in one concentrated part of the acquisition, it is challenging to take into account the affects of through-plane motion or motion that happens throughout the scan.

These limitations have required specialized developments for cardiac MRF (cMRF) [10], which will be discussed in this presentation. Firstly, due to the field inhomogeneities, a FISP-based [11] sequence is employed instead of the originally suggested bSSFP sequence. This leads to a reduced signal level but also relative insensitivity to off-resonance effects. Secondly, both cardiac and respiratory motion present challenges to cMRF due to the through-plane and continuous motion. In this implementation of cMRF, magnetization is excited and data are collected during a breathhold and only in diastole. In order to determine the signal time courses for pattern matching, the subject’s heart rate, and thus the exact timing of the scan, is used as an input into the Bloch simulation to determine a subject-specific dictionary. This solution leads to another set of difficulties, namely the “start-and-stop” nature of the scan, which results in T₁ recovery during the non-imaging periods. In order to maintain differences in signal evolutions for tissues with different T₁ and T₂ values, both inversion pulses and variable T₂-preparation modules are used during the scan. Because the scan time is limited to the length of a breathhold, the T_R is held constant at a minimal value such that the number of images collected can be maximized. Finally, the flip angle is varied but kept low to minimize effects from imperfect slice profiles and B₁⁺ inhomogeneities, which could lead to errors in the T₁ and T₂ maps. The resulting cMRF sequence is a 10- to 15-heartbeat scan performed in diastole where the following 5-heartbeat pattern is repeated (either two or three times): Inversion, no prep, T₂ prep (30ms), T₂ prep (50ms), T₂ prep (80ms). The spatial resolution is typically set to 1.6x1.6x8 mm³, with a matrix size of 192x192 and a scan window of 250ms. The cMRF sequence has been tested in phantom and normal volunteer studies, and the results for both T₁ and T₂ agree with literature values and maps made using standard cardiac relaxometry sequences (MOLLI and bSSFP T₂) at 1.5 and 3T [10]. An example set of maps from a normal volunteer is shown in Figure 1.

Several additional modifications to the cMRF technique have recently been proposed. The first is the use of a compressed sensing like reconstruction, known as Sparse MRF [12], instead of the basic inner product matching for the generation of parameter maps. This iterative reconstruction can reduce the number of images required for a given precision in the cMRF measurements; concretely, it can be used to reduce the number of heartbeats needed from 10 to 5, thereby reducing the total breathhold length (see Figure 2). Sparse MRF has also enabled the collection of time courses from several cardiac slices simultaneously in combination with SMS imaging. SMS cMRF has been used to collect T₁ and T₂ maps from 3-6 separate slices in 10 heartbeats [13], as shown in Figure 3.

In conclusion, MRF is an emerging technique for the simultaneous collecting of T₁ and T₂ maps. Cardiac MRF is also possible, and can be used to make multiparametric maps of one or more 2D slices in 5 to 10 heartbeats. Because the signal time courses are matched to relaxation parameters using direct Bloch equation simulations, changing heart rates and arrhythmias can be incorporated in the MRF dictionary, which may lead to fewer errors in the parameter maps than standard imaging sequences.

Acknowledgements

References