To optimize high resolution magnetic resonance vessel wall imaging regarding scan duration signal-to-noise and contrast-to-noise.

High resolution magnetic resonance imaging (MRI) plays an increasing role in diagnosing intracranial vascular diseases^1,2. By increasing the field strength a higher spatial resolution can be achieved and well-designed vessel wall sequences are developed to depict the intracranial vessel wall in further detail^2-7. Previously reported vessel wall sequences take 7-10 minutes^3,6,7. This time counts double as pre and post contrast scans are needed, which may makes it challenging to include other sequences required for examination, particularly in neurologically impaired patients. Ideally, the total scan time of the pre and post contrast vessel wall sequence should be reduced.
In this study we compared the signal-to-noise (SNR) and contrast-to-noise ratios (CNR) of two clinically used intracranial vessel wall sequences and 5 variants with various trade-offs between scan time, resolution and contrast between vessel wall and cerebrospinal fluid (CSF).Participant selection
MRI examinations were performed on five healthy subjects (4 male, aged 28 – 54 years). All subjects gave written informed consent.

Imaging protocol
A 3T MR system (Achieva 3.0T, Philips Healthcare, Best, The Netherlands) was used with an 8-channel phased array SENSE head coil. All vessel wall sequences consisted of a 3D turbo spin echo (TSE) sequence. The parameters form a multidimensional landscape, which cannot be fully explored in practice. Starting from two clinically used protocols, 5 variants were made, creating various tradeoffs between contrast (anti-DRIVE on/off for better CSF suppression), resolution (isotropic and nearly isotropic scans), SNR (by varying the oversample factor, SENSE factor, TSE train, and bandwidth), and scan duration. The scan parameters are summarized in Table 1. The total scan duration took approximately 1 hour, and included the preparation phase, a Time-Of-Flight MRA and consecutive the 7 vessel wall sequences. The vessel wall sequences were repeated without radio frequency (RF) pulse and gradients to sample the noise. These noise acquisitions were accelerated by removing the waiting time between end of the TSE train and the next excitation.

Data quantification and visualization
SNRs were calculated as an average of all subjects following SNR=Mean_ROI/SD_noise, where the means and standard deviations were derived from the magnitude images and noise images respectively. Regions of interest (ROIs) for the mean were manually drawn on a Philips Standalone Workstation from the circumferential of the basilar and left and right carotid artery, the CSF (suprasellar cistern), blood (medial cerebral artery) and frontal lobe (orbital gyri) as reference for brain tissue. ROIs of the standard deviation were drawn as a square circumferential of the ROI of the mean. The same locations for the ROIs were used within the subjects. CNRs were calculated as CNR_x-y=SNR_x-SNR_y.The quality of all images was sufficient to examine the SNRs and CNRs, which are displayed in Table 2. Images of the 7 sequences are also shown in Figure 1. Four of our tested sequences had shorter scan durations compared to previously proposed sequences^3,6. Our variant 7 had over 40% shorter scan duration compared to variant 2³. Moreover, variant 7 had comparable or even higher SNRs, and the highest CNR_{vessel wall-CSF} of all tested methods (Table 2). Especially in the elderly this may be beneficial, because with ageing brain atrophy will increase and the intracranial vessels are surrounded by more CSF. A disadvantage of variant 7 is the anisotropic resolution that is particularly visible on coronal and sagittal reconstructions. Yet, in clinical practice primarily axial images are used to examine the vessel wall. Because of the higher minimum refocusing angles in the TSE train of variant 2, 4, 5 and 7 they may yield more signal, but probably at the cost of blurring due to a worse pointspread function. However, these sequences do show the vessel wall clearly (Figure 1).
Due to flow effects of the CSF in the basal cisterns, a better delineation could be made of the proximal intracranial arteries at the level of the circle of Willis. In general, in this study the SNR of the vessel wall may be underestimated because of partial volume effect and difficulties of manually drawing a ROI around the vessel wall. This effect is even more obvious in the thinner basilar artery vessel wall, where the measured SNRs are consistently lower compared to the carotid artery vessel wall with surrounding tissue/CSF.We developed a considerable faster clinical feasible vessel wall sequence (4:38min) with high SNRs and CNRs that resulted in a good visibility of the intracranial vessel wall.