Background

Myocardial T₂ mapping is commonly used for the assessment of edema and advanced myocardial tissue characterization in cardiovascular magnetic resonance imaging (CMR).[1,2] Conventionally, T₂ mapping is performed by acquiring multiple T₂-weighted magnitude images and reconstructed by an exponential fit. T₂ preparation modules are most often used for generating these T₂ contrasts, requiring multiple seconds rest periods to achieve full magnetization recovery before each preparation. This leads to long-scan times and hampers clinical use. Recently, a new technique was developed to encode the T₂ information in the signal phase of a gradient-echo acquisition (GRE), circumventing the need for rest periods.[3] In this work we show the feasibility of phase-based T₂ mapping in the transient state in the myocardium in vivo and in a phantom.

Methods

The proposed method is based on the signal phase during the transient state of an RF phase modulated gradient echo (GRE) sequence with small quadratic phase increments (Figure 1A). Bloch simulations were performed to study the phase evolution as a function of T₁ and T₂ (Figure 1B,C,D). Slice profile effects were incorporated by simulating 100 isochromats across the slice thickness. Furthermore, B₁⁺ inhomogeneities were included by scaling the flip angle. Look-up tables were generated for T₂ map reconstruction containing a variety of B₁⁺ values (0.6-1.2 of the nominal flip angle) to assign each voxel a B₁⁺ dependent T₂ estimate.
In the proposed sequence scheme five slices were acquired using a GRE sequence with positive phase increment Φ=2°, followed by five slices with negative phase increment Φ=-2° (Figure 1F). The two images for each slice were subtracted before T₂ mapping in order to eliminate the background phase. All images were acquired at 3T (Magnetom Skyra, Siemens Healthineers). Phantom measurements were performed in 13 tubes with T₁ times all above 2000 ms and T₂ times varying between 30 ms and 200 ms to separately analyze the dependency in T₂. Additionally, accuracy and precision was validated in the phantom and correlated to the Bloch simulations for the T₂ dependency. Bloch-Siegert mapping was acquired in the same five slices to correct for B₁⁺ inhomogeneities.
In vivo measurements were performed during end diastole in two healthy volunteers with the same sequence parameters as for the phantom. The estimated phase based T₂ maps are compared with the reference cardiac T₂ method[4].
Sequence parameters were: TE=1.8 ms, TR=3.3 ms, flip angle α=15°, FOV = 384x288 mm², in-plane resolution of 1.0x1.0 mm², acquired lines to the center of k-space = 61 and switching phase increments of ±2°.

Results

Simulations show that the signal phase is strongly dependent on T₂ but not on T₁ and asymmetric for positive and negative phase increments. The subtracted phase (labeled as phase difference) shows, therefore, a major dependence on T₂ but negligible T₁ effects (Figure 1E). During the transient state strong B₁⁺ dependence is observed.
Phantom measurements agree with the simulations and show a major T₂ dependence of the phase difference (Figure 2). In comparison to phantom images, simulations overestimate the phase difference when no correction for B₁⁺ and slice profile are applied (gray dotted line). The overestimation was reduced when incorporating slice profile effects into the simulations (black dotted line). Resulting T₂ times are in good correlation (R=0.95, p<0.001) when additionally B₁⁺ was included (Figure 2B). The average B₁⁺ value in the tubes was 0.96 resulting in an effective flip angle of 14.4° (Figure 2A). Visually good agreement between the spin-echo reference method and phase-based T₂ mapping is achieved (Figure 2C). T₂ times estimated in phantom were within 6% deviation compared with the reference for T₂ less than 100 ms. Higher deviations were observed at very long T₂ times (Figure 2D).
In vivo phase based T₂ mapping yielded mean T₂ times of 29.9 ms, which is 11% lower than the reference T₂ map 33.6 ms. The myocardium appears homogeneous, especially for the mid ventricular slice, and results in visual image quality comparable to the reference method (Figure 3). Residual B₁⁺ effects were observed in the septal wall.

Discussion

We demonstrated the feasibility of using transient state phase based T₂ mapping for fast T₂ quantification in the myocardium, obtaining good correlation between simulations and phantom measurements and satisfactory image quality in vivo. Phase based T₂ mapping enables the quantification of five slices in one breath-hold (10 heart-beats), which is similar to the reference T₂ prepared single-shot acquisition covering only one slice. This enables whole heart T₂ mapping within one breath-hold. Phase based T₂ mapping does not provide blood T₂ times because of the phase cancellations. Furthermore, residual subject specific B₁⁺ is observed, which needs to be corrected for. Further improvements might be joint quantification of B₁⁺ and T₂ and warrant further investigation.

Conclusion

Myocardial phase based T₂ mapping is a promising new technique for fast quantification of multiple slices in the myocardium.

Acknowledgements

S.W. acknowledges funding from the 4TU Precision Medicine program, a NWO Start-up STU.019.024 and ZonMW OffRoad 04510011910073.