Carotid plaque assessment using fast 3D isotropic resolution black-blood MRI

Abstract

Black-blood MRI is a promising tool for carotid atherosclerotic plaque burden assessment and compositional analysis. However, current sequences are limited by large slice thickness. Accuracy of measurement can be improved by moving to isotropic imaging but can be challenging for patient compliance due to long scan times. We present a fast isotropic high spatial resolution (0.7×0.7×0.7 mm3) three-dimensional black-blood sequence (3D-MERGE) covering the entire cervical carotid arteries within 2 min thus ensuring patient compliance and diagnostic image quality. The sequence is optimized for vessel wall imaging of the carotid bifurcation based on its signal properties. The optimized sequence is validated on patients with significant carotid plaque. Quantitative plaque morphology measurements and signal-to-noise ratio measures show that 3D-MERGE provides good blood suppression and comparable plaque burden measurements to existing MRI protocols. 3D-MERGE is a promising new tool for fast and accurate plaque burden assessment in patients with atherosclerotic plaque.

Figure 1

Sequence diagram of 3D-MERGE: An MSDE black-blood preparation is followed by fat saturation pulse and N low flip-angle spoiled gradient echo acquisitions with centric encoding without preparatory RF pulses.

Figure 2

MSDE prepared snapshot FLASH signal (a) compared to snapshot FLASH signal without magnetization preparation (b) for muscle tissue with T1=1000ms, T2=40ms. Bloch simulation parameters matched the invivo protocol used (Table 1): TR-10ms, TE=4.8ms, T_m-22.5ms, T_d-12.25ms. The total signal (solid line) is reduced by the MSDE preparation at short turbo factors compared to FLASH without preparation. While the relaxation due to the gradient echo readout (dotted line) is relatively unaffected, the transient component (dashed line in 2b) of snapshot FLASH is split into two components by the MSDE preparation. The component due to post-MSDE relaxation is negligible (dot-dash line) while the component due to T2 relaxation (dotted line in 2a) is the primary signal component during the approach to steady state at short turbo factors and is determined by T_m (see figure 1). For the MSDE first gradient moment (1574 mT ms² /m) and gradient strength (20mT/m) used for invivo imaging T_m was 22.5ms. For these parameters, increasing turbo factors increases signal intensity. The effect is more pronounced for tissues with shorter T1 due to faster T1 recovery (c). For all T1, signal intensity can be improved by reducing T_m. (d) shows the increase in M_z at shorter turbo factors when T_m is reduced by one-thirds compared to (c).

Figure 3

Dependence of blood suppression in 3D-MERGE on Turbo factor. Signal due to blood (T1=1550ms, T2=275ms) recovers after each MSDE prepulse with a modulation of k-space that depends on the acquisition turbo factor. Simulation shows lower blood signal with short turbo factor (30) compared to longer turbo factors (90,120). With long turbo factors signal recovery is faster with increasing k_y. Shorter turbo factors require more MSDE shots to cover k-space thereby improving blood suppression. Recovery of blood longitudinal magnetization is also smaller with short turbo factors thereby contributing to improved blood suppression. Increased luminal signal with increasing turbo factor can be seen in figure 4

Figure 4

Overall signal can be seen to improve with increasing turbo factor (a) in images of a volunteer. The luminal signal also increases (arrow) due to the longitudinal relaxation of blood and the reduced number of central k-space lines acquired immediately after an MSDE pulse with increasing turbo factors. Reducing T_m can improve signal as predicted in figure 2d. With a one-third reduction in MSDE m₁ from 1574 mT ms² /m to 320 mT ms² /m and reducing turbo factor to 30, higher signal than (a) can be achieved. However, note the appearance of slight plaque mimicking artifacts in (b) near the lumen-vessel wall boundary. All images are presented with the same window-level.

Figure 5

Reformats of isotropic 3D-MERGE in all three orthogonal planes showing good vessel wall delineation in a patient. Calcification of right carotid plaque is clearly visible (arrow).

Figure 6

Axial reformat (a) of bilateral carotid plaque (arrows) in a patient with a large plaque in the left carotid artery. Sagittal reformats of right (b) and left (c) carotid arteries show the extent of plaque (arrows) along the length of the artery.

Figure 7

Representative images from a patient: (a) PDw 2mm slice thickness, (b) 3D-MERGE reformatted axially to match (a) with 2mm slice thickness and (c) Five slices of axial 3D-MERGE (0.4mm thick) corresponding to (a) and (b). Visualization of small calcification (arrow) can also be seen to be impaired due to partial volume averaging in (b).

Figure 8

Three consecutive slices showing calcified plaque in a patient. PDw 2mm (first column) and 3D-MERGE 2mm show calcifications (arrow) with better visualization on 3D-MERGE. Corresponding slices of 3D-MERGE 0.4mm show the change in shape of calcium from slice to slice which can be useful in detecting small juxta-luminal calcifications.

Figure 9

Bland-Altman plots comparing morphological measurements between 3D-MERGE 2mm and PDw 2mm. Solid and dotted lines correspond to mean difference and limits of agreement, respectively. LA: Lumen Area, WA: Wall Area, MWT: Mean Wall Thickness, MaxWT: Maximum Wall Thickness.

Figure 10

Bland-Altman plots comparing morphological measurements between 3D-MERGE 0.4mm and PDw 2mm. Solid and dotted lines correspond to mean difference and limits of agreement, respectively. LA: Lumen Area, WA: Wall Area, MWT: Mean Wall Thickness, MaxWT: Maximum Wall Thickness.

Figure 11

Correlation of lumen and wall area measurements between (a) 3D-MERGE 2mm and PDw 2mm (b) 3D-MERGE 0.4mm and PDw 2mm. Solid and dotted lines are regression line and 95% confidence intervals, respectively.