Fig 1.
a) Pulse sequence diagrams for the conventional adiabatic T2-Prep and b) the 2D-T2-Prep. In the 2D-T2-Prep, the initial tip-down pulse has been replaced by a spatially selective jinc pulse, which excites a cylindrical volume (12.0 cm). The final pulse, which remains non-selective, restores this T2-Prepared cylinder while tipping outer volume magnetization into the transverse plane, where it is spoiled. Both of the effective T2-Prep durations remain the same and either may be used prior imaging. c,d) A 50 scan average of both techniques. In these images, a homebuilt phantom, with compartments doped to mimic blood, fat, and myocardium, was placed in a large water bath and imaged with a 2D sequence preceded by either the c) conventional T2-Prep or the d) 2D-T2-Prep.
Fig 2.
Numerically simulated GRAPPA-accelerated acquisitions of a phantom.
Random coil noise was added to the 50 scan average described in Fig 1C) for the conventional T2-Prep as well as for 2D-T2-Prep modules of radius 6.0 and 12.0 cm. GRAPPA-accelerated image acquisitions were simulated for acceleration factors of R = 1 to 6. Sample images can be seen in a), with corresponding per-pixel maps of b) SNR, c) Noise, and d) G-factor. Note that accelerated image quality appears to improve as the excitation volume decreases, though the position and nature of GRAPPA artefacts also appear to shift.
Fig 3.
Improvement in the mean SNRmulti, in regions of a tissue-mimicking phantom, when using a 2D-T2-Prep instead of a conventional T2-Prep, for GRAPPA accelerations of R = 1,2,3,4,5, and 6.
Results are shown for both a) the simulated, ideal T2-Prep shown in Fig 2 and b) the corresponding physical phantom. Fig a) demonstrates the maximum possible signal improvement achievable, in the case of perfect background suppression with a 6.0 cm radius cylinder, whereas Fig b) demonstrates the actual results achieved in the physical phantom. In both cases the maximal SNR improvement appears to occur at R = 4, which also corresponds to the degree of outer volume suppression of the 2D-T2-Prep.
Fig 4.
Sample scan (R = 1) of the phantom used for vessel sharpness measurements, imaged using a) the conventional adiabatic T2-Prep and b) the 2D-T2-Prep. Images were acquired with acceleration factors of R = 1,2,3,4,5, and 6, where the phase encoded (accelerated) direction is indicated by the two-headed arrow at the bottom right. Vessel sharpness was measured along the artificial vessel, in portions parallel to the accelerated direction (solid arrow) and perpendicular to the accelerated direction (dashed arrow).
Fig 5.
Improvement in the mean vessel sharpness, as measured in a coronary-mimicking phantom, when using a 2D-T2-Prep instead of a conventional T2-Prep, for GRAPPA accelerations of R = 1,2,3,4,5, and 6.
Results are shown for a "vessel" section a) parallel to the accelerated direction (gray triangles) and b) perpendicular to the accelerated direction (black squares). As artefacts propagate along the accelerated direction, the parallel case acts as a vessel sharpness control, whereas the perpendicular case may be used to determine the relative benefit of the 2D-T2-Prep.
Fig 6.
Sample in vivo images of the RCA at various GRAPPA-acceleration factors, for both a-c) the conventional adiabatic T2-Prep and d-f) the 2D-T2-Prep. The 2D-T2-Prep increased the mean vessel sharpness, as compared to the conventional T2-Prep, even in the non-accelerated case. However, this relative improvement was greater at R = 3 and greater still at R = 6, consistent with the phantom data shown in Fig 5. Note the significantly decreased conspicuity of the distal RCA, particularly for the R = 6 case, when using the conventional T2-Prep. Conversely, this is still quite visible with the 2D-T2-Prep (white arrows).
Table 1.
Improvement in RCA Vessel Sharpness using the 2D-T2-Prep for R = 1,3, and 6.