3D Carotid Wall Imaging: Stack-of-stars Trajectory for Multi-contrast Atherosclerosis Characterization (STAR-MATCH)
Xiaoming Bi1, Zhaoyang Fan2, Yutaka Natsuaki1, Debiao Li2, and Gerhard Laub1

1Siemens Healthcare, Los Angeles, CA, United States, 2Cedars-Sinai Medical Center, Los Angeles, CA, United States


The recently developed MATCH technique integrates multiple 3D image sets into a single measurement and it is a promising method for carotid plaque characterization. One of the remaining challenges is the gross motion of carotid arteries that originates from pulsation, breathing and swallowing. In this work, a motion robust stack-of-stars sampling trajectory was implemented into the MATCH sequence (STAR-MATCH). Preliminary studies from volunteers and patient demonstrate it is feasible to characterize carotid plaque using the STAR-MATCH sequence with improve motion robustness.


2D Multi-contrast MR imaging has been well established for the carotid plaque characterization [1-3]. Such method, however, has limited slice resolution, spatial coverage, and potential mis-registration between different images. The recently developed MATCH technique integrates multiple 3D image sets into a single measurement [4]. Plaque component-specific contrasts setting and intrinsic registration of multiple images acquired with MATCH greatly simplify the workflow and image interpretation. One of the remaining challenges for the MATCH technique is the gross motion of carotid arteries (e.g. pulsation, breathing and swallowing) that compromises the vessel wall delineation [5]. Stack-of-stars sampling scheme was recently demonstrated to have excellent tolerance to carotid motion [6]. We hypothesize that MATCH sequence benefits from integrating the stack-of-stars sampling trajectory (i.e. STAR-MATCH). The current work investigates feasibility of using STAR-MATCH sequence for carotid plaque characterization.


Figure 1 shows the schematic diagram of the STAR-MATCH sequence. Following a non-selective inversion preparation, three image sets are acquired at different TI time points using stack-of-stars sampling trajectory. Readouts for image sets 1 and 3 are preceded by a DANTE preparation [7] module for blood signal suppression. TI values and magnetization preparations are optimized to delineate short-T1 hemorrhage, calcification, and overall plaque morphology, respectively, in three image sets. Three healthy subjects and one patient with carotid plaque were scanned on a 3T MR scanner (MAGNETOM Prisma, Siemens Healthcare, Erlangen, Germany). An eight-channel carotid coil (4 elements on each side) was used as signal receiver in combination with a 20-channel head-neck coil. For the comparison purpose, MATCH sequence using Cartesian sampling was also run on same volunteers with identical protocol settings as with STAR-MATCH. Imaging parameters included: FOV = 16 x 16 cm2; 64 coronal partitions; acquisition voxel = 0.83 x 0.83 x 0.83 mm3; flip angle = 8°; 192 radial views (STAR-MATCH) or phase-encoding lines (MATCH); TI1/TI2/TI3 = 370/1200/2650 ms; repetition time between inversion pulse (IRTR) = 3000 ms. DANTE module (150 ms) was composed of 120 hard RF pulse (FA = 12°) with spoiler gradients (20 mT/m, 0.9 ms duration) in between. For the patient scan, 3D TOF, 2D T1-weighted and T2-weighted TSE images were also acquired. Parameters for TSE included: 5 axial slices; 0.51 x 0.51 x 3.00 mm3 voxel size.


Carotid artery images were successfully acquired from all subjects. Figure 2 shows representative images acquired from a healthy subject. Good quality images were acquired with three desired tissue contrasts using both MATCH and STAR-MATCH sequences. Wrapping artifacts in Cartesian acquisition were not present in STAR-MATCH images. MIP of STAR-MATCH morphological image (image 3) showed improved carotid wall delineation using STAR-MATCH. Figure 3 shows MPR images of patient scanned with STAR-MATCH and conventional TOF, 2D TSE sequences. No intraplaque hemorrhage was detected from the Hyper-T1 (image 2) STAR-MATCH images. As illustrated in Figure 4, focal signal voids were detected in the gray-blood contrast (image 2) presumably due to calcification of local vessel wall. Ulcerated plaque is clearly visible in the sagittal MIP of 3rd STAR-MATCH image. Axial MIPs of STAR-MATCH agree corresponding TSE and TOF images acquired from the same slice location. Figure 5 illustrates sagittal MPRs of STAR-MATCH and TOF. Fused TOF and STAR-MATCH image demonstrates excellent spatial registration between these two images. Atheroma and surrounding blood can be clearly differentiated. In comparison, 2D TSE shows poor slice resolution and spatial coverage despite long imaging time (4.5 minutes for T1-TSE).


Taking advantage of radial sampling scheme, STAR-MATCH offers improved robustness to motion compared to its Cartesian counterpart. This can be used to improve spatial coverage and/or resolution despite slightly longer imaging time. Future work on combining STAR-MATCH with MR acceleration technique (e.g. parallel imaging or other advanced reconstruction) is warranted. Also its performance in patient needs to be evaluated in a larger study population.


STAR-MATCH is a promising 3D technique for the characterization of carotid plaque in a single all-inclusive measurement with improved carotid plaque motion robustness.


No acknowledgement found.


[1] Yuan C et al, Circulation 2001, p2051.

[2] Fayad ZA et al, Ann NY Acad Sci. 2000, p173.

[3] Saam T et al, Arterioscler Thromb Vasc Biol, 2005, p234.

[4] Fan Z et al, JCMR 2014, p53.

[5] Boussel L et al, JMRI 2006, p413.

[6] Bi X et al, ISMRM 2015, p555.

[7] Li L et al, MRM 2012, p1423


Figure 1. Schematic diagram of the STAR-MATCH sequence for carotid wall imaging. Three images are acquired using stack-of-stars readout at different TI values. Imaging contrasts are optimized for assessing hemorrhage (Hyper-T1), calcification (Gray blood) and plaque burden. Non-selective inversion preparation is repeatedly applied for every 3000 ms (IRTR). Within each IRTR, same radial view is acquired for all partitions of three image sets (red radial view as an example).

Figure 2. Representative volunteer images acquired with MATCH (top row, Cartesian sampling) and the proposed STAR-MATCH (bottom row). Desirable image contrast was achieved for all images. Note that wrapping artifacts (ovals) using Cartesian sampling are not present in STAR-MATCH images. Also MIP of image 3 shows improved delineation of carotid wall (arrows) using STAR-MATCH sequence.

Figure 3. MIP images acquired from a 66 years old patient using STAR-MATCH and the conventional TOF & 2D TSE sequences. Ulceration of the plaque was clearly visible in the sagittal MIP image. Axial MPR images of STAR-MATCH from two slices agreed with conventional TSE and TOF images acquired at the same slice.

Figure 4. Sagittal (top row) and axial (bottom) MIP images acquired from a 66 years old patient using STAR-MATCH sequence. From gray blood images (left column), focal signal voids were detected (solid circles) presumably due to local calcification of vessel wall. Corresponding positions in dark-blood images were illustrated (dotted circles in right column).

Figure 5. MPR of STAR-MATCH showed large spatial coverage and good delineation of carotid wall. Corresponding TOF image showed signal loss (red arrow) distal to the plaque presumably due to local turbulent. Blood suppression in the same region was excellent using STAR-MATCH (green arrow). Fused TOF and STAR-MATCH image demonstrated nice spatial registration between them. MPR of 2D TSE images had poor slice resolution and limited spatial coverage despite long imaging time (4.5 mins).

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)