MRI of the carotid arteries has emerged as an important imaging field during the last decade and has been shown valuable for risk assessment of plaque rupture.¹ The use of surface coils for the neck is mandatory in order to optimize the SNR performance. Parallel imaging does even further increase the performance of carotid MRI since several applications are nowadays hampered by excessive scan duration and limited temporal resolution. Especially 4D flow measurements in the carotids at sufficiently high spatiotemporal resolution² (<1mm,~50 ms) have not been proven feasible to date. We hypothesize that a dedicated 15 channel carotid surface coil allows for 10 fold accelerated 4D flow by application of parallel imaging in the two phase encoding directions.

Two different experiments were conducted in two healthy volunteers (1 male). We compared SNR and g-factor performance between a conventional 8 channel bilateral coil (Shanghai Chenguang Medical Technologies, Shanghai, China) and a novel 15 channel unilateral coil (MR Coils, Drunen, The Netherlands). We subsequently present a show case to prove the performance of the novel 15 channel coil for 4D flow in the carotids. The 8 channel coil has two coils with 4 coil elements in a 2x2 configuration. The 15 channel coil has a 3x5 configuration, as described by Koning et al.³, with 5 coils in the anterior-posterior direction. All data was acquired on a 3T MRI scanner (Philips Ingenia, Best, the Netherlands). To compare SNR, a multi slice FFE image stack and the corresponding noise images of the right carotid artery were acquired using both coils; scan parameters were: FOV 200x200 mm, TR/TE/FA 9.1/4.2/6, 10 slices, 8 mm gap, resolution 1x1 mm, 1 mm slice thickness. The SNR maps of the 8 channel and 15 channel coils were compared by plotting the SNR along a manually drawn line at equal positions in the neck.

To compare the parallel imaging capability of the 15 channel coil, we acquired multiple single-slice TSE images with increasing parallel imaging factors (SENSE 2,3 and 4) in the AP-direction and g-factor maps were reconstructed on the scanner. Scan parameters were FOV 150x150mm, TR/TE/FA 800/11/90, ETL 10, resolution 0.28x0.28 mm, 3 mm slice thickness. In addition, the noise correlation between the 15 elements was assessed in vivo. The loaded and unloaded Q factor was determined to assess the noise factor of the 15 channel array.

4D flow was measured using the following sequence parameters: 4 point velocity encoding, TR/TE 7.6/3.7 ms, Venc (RL,AP,FH) 60/60/100 cm/s, FOV (RL,AP,FH) 215x129x90 cm, FA 15 degrees, non interpolated resolution (RL,AP,FH) 0.9x0.9x1.4 mm, temporal resolution 61 ms, SENSE (AP, FH) 4x2.5 (acceleration factor 10), scantime 7.14 min. For comparison a 2D flow scan was acquired in the common carotid artery at high temporal resolution of 18 ms. G-factor maps were reconstructed for SENSE 4x2.5.

The SNR of the 15 channel coil was 1.5 times higher at 1 cm depths, whereas at the location of carotid arteries the SNR was comparable (figure 1A-C). The g-factor maps measured in a healthy volunteer revealed that with increasing SENSE the g-factor loss in the 15 channel coil is substantially less compared to the 8 channel coil (figure 2). In the magnitude images (figure 2, gray images) a better SNR at 3 cm depth for SENSE 4 could indeed be observed for the 15 channel coil, in agreement with the lower g-factor maps. The cross correlation matrix (figure 1D) shows that the average cross correlation was limited to ~12%. The ratio between unloaded Q factor (348) and loaded Q factor (97) is 3.6 fold, demonstrating tissue loss dominance (and a low noise figure).

In the second experiment 4D flow was successfully performed with large coverage at high spatiotemporal resolution, as can be appreciated in the streamline visualization infFigure 3. The flow curve over the heartcycle is still similar to the flow curve obtained at a 3-fold higher temporal resolution and the peak flow is captured correctly (figure 4). G-factors remained below 1.5 at the location of the carotid artery (figure 5).

Using a novel 15 channel coil we could prove that the g-factor loss at higher acceleration factors is limited. This permits the use of higher parallel imaging factors than commonly exploited in carotid MRI.4D flow at sufficiently high spatiotemporal resolution in the carotid arteries becomes feasible for the first time using a dedicated 15 channel coil. Once a bilateral 30 channel design becomes available in the near future this coil will allow for routine 4D flow assessment of the carotid arteries within feasible scantime.