High resolution carotid black-blood 3T MR with parallel imaging and dedicated 4-channel surface coils

Complications of cardiovascular disease, including stroke, myocardial infarction, and sudden cardiac death, are the most common causes of death in the western world. The challenge for screening and diagnostic methods is to identify patients at high risk who have lesions that are vulnerable to thrombosis, so-called "vulnerable plaques", before the event occurs. Imaging methods have the potential to not only be used as a screening tool for the presence of atherosclerosis but also to help distinguish stable from vulnerable plaques and finally to distinguish between patients with low risk from those with high risk of cardiovascular complications[1].

Traditionally, atherosclerosis imaging was focused on the depiction of the arterial lumen[2]. However, atherosclerosis is a disease which begins in the arterial wall and hemodynamically relevant luminal narrowing potentially occurs rather late in the atherosclerotic disease process. Therefore, it is now commonly accepted[3, 4] that knowledge of the luminal diameter is not sufficient to determine the vulnerability of an atherosclerotic lesion, and a number of major and minor criteria have been proposed for the assessment of a vulnerable plaque. These plaque features were based on studies of coronary arteries and included thin caps with large lipid/necrotic core, active inflammation, fissured plaque, stenosis >90%, endothelial denudation with or without superficial platelet aggregation and fibrin deposition, endothelial dysfunction, calcified nodules, intraplaque hemorrhage, glistening yellow plaques (on angioscopy), and outward remodeling. Hence, the ideal atherosclerosis imaging method should not only be able to quantify the degree of luminal stenosis but also to identify these key features defining a vulnerable plaque.

Although several imaging modalities, such as intravascular ultrasound (IVUS)[5] or computed tomography[6], have been in use to assess atherosclerotic plaques, magnetic resonance (MR) provides the unique potential to identify most of the key features of the vulnerable carotid plaque[1]. MR is non-invasive, does not involve ionizing radiation, enables the visualization of the vessel lumen and wall[7] and can be repeated serially to track progression or regression[8]. Furthermore, the excellent soft tissue contrast provided by MR enables the evaluation of compositional and morphological features of atherosclerotic plaques[1].

Recent MR studies. [8‐12] have shown that this method could become a useful clinical tool to assess the carotid atherosclerotic plaque. However, many of these carotid plaque MR studies. [8‐10] have been performed using black-blood protocols at 1.5 T without parallel imaging techniques. Therefore, disadvantages of previous carotid MR studies at 1.5 T were long scan times of up to 45 minutes and a relatively high number of excluded subjects due to insufficient image quality in 5-33% of the subjects[13]. Furthermore, the spatial resolution of 0.6 × 0.6 mm typically used for 1.5 T MR studies might not be able to identify very small features of the atherosclerotic plaque, such as the fibrous cap.

Recently, clinical MR at 3 T, which can potentially increase the signal obtained during imaging by a factor of 2, has become widely available and several carotid black-blood 3 T MR studies [11, 12, 14, 15] have shown that the added signal can be used to either decrease scan time or to reduce spatial resolution or a combination of both. However, these carotid MR studies [11, 12, 14, 15] did not use parallel imaging techniques. Parallel imaging techniques are being used increasingly in clinical practice [16] and have been used recently for aortic atherosclerosis imaging using a sensitivity encoding (SENSE) algorithm [17]. Briefly, parallel acquisition techniques combine the signals of several coil elements in a phased array to reconstruct the image, the chief objective being either to improve the signal-to-noise ratio or to accelerate acquisition and reduce scan time. Shorter scan times could potentially reduce motion artefacts caused by respiration and swallowing.

To further improve carotid black-blood imaging and to facilitate its usefulness in routine clinical practice we propose a multi-sequence carotid black-blood protocol using parallel imaging and a dedicated 4-channel surface coil on a 3 T MR scanner, to reduce acquisition time per slice by factor 2, minimize dropouts and optimize overall image quality. To evaluate the clinical usefulness of our protocol we evaluated the reproducibility of quantitative plaque measurements and we compared compositional and morphological plaque features between healthy volunteers and subjects with carotid atherosclerotic disease.

The results of this study demonstrate that high-resolution 3 T carotid MR with parallel imaging has a high reproducibility for quantitative measurements of lumen area, wall area, total vessel area and NWI. With our standardized protocol, implementing parallel imaging techniques and dedicated surface coils, we achieved sufficient image quality and coverage to perform quantitative measurements in 37 out of 38 scans and good intra-scan reproducibility was obtained across the two different time points in the 10 volunteers which were measured twice, optimizing the co-registration by using the bifurcation as a landmark (figure 1). Furthermore, this study was able to show highly significant differences of quantitative and qualitative plaque measurements between the 14 volunteers and the 13 patients, demonstrating that MR is able to detect significant differences between populations, even in small patient cohorts.

Similarly to previous studies on 1.5 T [22, 27], and 3 T MR systems [11, 15, 28] this study showed excellent correlation in the 10 volunteers for the baseline measurements compared to the second time point for all measured variables. This indicates that the normal variation between subjects of any of the measurements was much greater than the variation within the repeated measurements and expresses almost perfect agreement on repeated measurements (perfect agreement = 1.00). In contrast to previous studies [11, 15, 22, 27, 28] this was achieved by using parallel imaging techniques, which accelerated the image acquisition per slice by a factor of 2 for TOF, PDw, T2w and T1w images (see table 3),

The measurement error or modified coefficient of variation ranged from 1.5-4.5% for mean lumen area, mean wall area, mean total vessel area and NWI and was slightly lower than previously reported in 1.5 T MR studies (see table 3 for comparison of 1.5 T and 3 T MR studies) in single [29] and multi-center trials [22] (5-6.8% for mean wall area measurements). Possible reasons for the lower measurement error reported in the current study might be due to the following reasons: i.) improvement of the spatial resolution which resulted in a 31% smaller voxel size compared to previous 1.5 T MR studies, ii.) use of parallel imaging techniques which resulted in a decrease of scan time and thus less time for possible patient movement and iii.) signal gains associated with the higher field strength which might have resulted in an improved lumen and wall boundary detection.

Due to the low variability of quantitative measurements, MR is able to detect differences between groups even in small study populations. This study was able to show significant differences in the CCA and/or ICA between 13 patients and 14 volunteers for all quantitative measurements and for the prevalence of features of advanced atherosclerotic disease, such as lipid-rich necrotic core, calcification or hemorrhage. As expected, patients had larger mean wall areas and mean NWI's in both CCA and ICA, which confirms the larger atherosclerotic plaque burden in patients. Total vessel area in the CCA was significantly larger in patients compared to volunteers, which indicates that the CCA in the patients had undergone outward remodeling, while mean lumen area in the CCA was preserved

This study found positive correlations with mean wall area and NWI in the CCA and/or ICA with age and all cardiovascular risk factors except family history of cardiovascular disease. This finding indicates that both NWI and mean wall area might be useful to assess the atherosclerotic disease burden. On the other side, mean lumen area failed to show significant correlations with cardiovascular risk factors except with hypercholesterolemia in the ICA, suggesting that assessment of mean lumen area is a less sensitive marker of the atherosclerotic disease burden than NWI and mean wall area. Interestingly, correlation coefficients of NWI with age and risk factors tended to be slightly larger than correlation coefficients of wall area with age and risk factors, suggesting that NWI might be the most sensitive measure to evaluate the severity of the atherosclerotic disease process.

In contrast to previous 1.5 T and 3.0-T MR studies [11, 15, 22, 27, 28], the 3 T MR study protocol presented here used parallel imaging techniques to image the bilateral carotid arteries. The technique was based on the generalized autocalibrating partially parallel acquisition (GRAPPA) algorithm [18] with a PAT factor of 2 for the TOF, T1w, T2w and PDw images and a PAT factor of 4 for the dynamic contrast enhanced sequences. This was associated with a substantially shorter scan time per slice which decreased between 47% and 59% for T1w, PDw/T2w and TOF images (see table 3). In order to ensure full coverage of the carotid atherosclerotic plaque bilaterally, the scan range was expanded from 24 to 30 mm. Thus, despite the increased coverage this protocol resulted in a total decrease of the scan time of 33% for non-contrast enhanced exams and 29% for contrast-enhanced exams.

A recent 3 T carotid MR study[28] has shown significant improvement in SNR, CNR, and image quality for high- resolution black-blood imaging of carotid arteries at 3 T compared to 1.5 T with a 1.5-fold SNR gain for T1-weighted images and a 1.7/1.8-fold gain for PD-/T2-weighted images. Another study [15] showed signal gains for carotid artery wall SNR at 3.0 T relative to 1.5 T of 223% and wall-lumen CNR of 255% in all acquisitions (P < 0.025). The signal gain associated with the higher field strength and the shorter scan time due to the parallel imaging technique was used to establish a clinical MR protocol which was faster, had a higher longitudinal coverage and a better spatial resolution than protocols previously used on 1.5 T MR scanners (see table 3) [22, 27]. This study achieved diagnostic image quality in 37 out of 38 scans, demonstrating that this protocol can be successfully used in a clinical setting. This drop-out rate is substantially lower than the drop-out rates reported in recent 1.5 T MR studies[20, 22, 30], in which 27 out of 252 (10.7%) exams were excluded due to insufficient image quality. It is also substantially lower than the drop-out rate reported in a recent multi-center MR study [13] in which 160 patients with greater than 50% carotid artery stenosis were recruited at six centers for prospective imaging of the carotid arteries at baseline and 1 year later by using high-spatial-resolution, 1.5 T MR imaging. All 160 patients completed both baseline and follow-up studies. Of these studies, only 67.5% were deemed to have image quality that was acceptable for quantitative analysis. The causes of rejection were motion (46%), deep location of the carotid artery (22%), low bifurcation of the carotid artery (13%), and "other" (19%). One of the reasons of the lower number of excluded exams in our study compared to previous 1.5 T MR studies might be the signal gains associated with the higher field strength in combination with the shorter scan time due to the Parallel Imaging technique, which is easier to tolerate for the patients and might therefore result in less motion artefacts and artefacts due to patient swallowing.

Recent 3 T carotid MR studies [14, 15] used coil designs in which the carotid coil could not be combined with other coils, such as head or neck coils. In contrast, this study used a dedicated 4-channel surface coil, which can be easily combined with all other commercially available coils for the 3 T imager used in this study. This is a substantial advantage over other coil designs as with this approach, an MR examination of the carotid arterial wall and a brain MR and/or an angiography of the supraaortal vessels can be performed without a coil change or patient repositioning.