Dynamic contrast enhanced (DCE) magnetic resonance angiography (MRA) is an established method for evaluating pathology in the thoracic aorta and supra-aortic vessels as well as arterial pathology in the neck. Good quality DCE MRA of the chest and neck is challenged by the need of high spatial resolution with sub-millimeter voxels, high temporal resolution to avoid venous contamination, and good suppression of the background signal to aid increased vessel conspicuity. Background suppression is generally achieved by subtracting a pre-contrast mask from the contrast enhanced image, utilizing 3D sequences with high flip angles and short TRs, such that the static tissues experience high level of T₁-based saturation, or a combination of both. While these approaches often result in high quality images, subtraction errors can sometimes lead to incomplete suppression of the fat due to motion, thus diminishing image quality and visualization of the vessels of interest. B₀ inhomogeneities resulting from air-tissue interfaces on both sides of the neck and the fast nature of the timed acquisition following the bolus injection limits the utility of frequency-selective fat-suppression techniques. It has been previously shown that a ‘subtraction-less’ MRA of the extremities using Dixon-based acquisitions to eliminate the background fat signal is a feasible and robust approach to generate high-quality MRA examinations¹. We investigated the ability of DCE dual-echo multi-peak Dixon-based imaging with flexible echo times² to generate MR angiography of the chest and neck, without the need of subtraction.

Patients and MRI Protocol: This was a prospective, IRB-approved, HIPAA-compliant study. Six patients with multiple-sclerosis (MS) scheduled for MRI of the brain and spine as part of their clinical evaluation agreed to participate and signed the informed consent. Patients were scanned using a 3T dual-transmit MRI with a head-neck neurovascular coil (Ingenia, Philips Healthcare, Best, the Netherlands). A thick-slab, low-resolution phase-contrast angiography image was first acquired to aid positioning of the 3D MRA volume. This was followed by a pre-contrast free-breathing oblique coronal three-dimensional (3D) elliptical-centric-ordered T1-weighted (T1W) fast gradient-echo (FFE) dual-echo 7-peak DIXON with correction for B₀ variations in large fields of view (mDIXON XD), with TR/TE = 5.4/1.93 ms, ΔTE = 1.47ms, FA = 270, resolution = 0.9×0.9×1 mm³ (acquired as 2mm overcontiguous), acquisition FOV = 400×300×73 mm³, (FH×LR×AP), acquisition matrix = 444×332×73, bandwidth/pixel = 853 Hz, acquisition time = 22 seconds/dynamic, SENSE = 3 right-left. In 5 cases no shimming was employed, while in one case a first-order automatic shimming was centered over the chest region. The arrival of the contrast bolus (0.1 mmol/kg Gadovist, 1.2 mL/s injection followed by 20-ml saline flush) was timed by a BolusTrak, followed by a repeated mDIXON acquisition with the same parameters.

Image Analysis: For each mDIXON acquisition, water-only images were reconstructed using a 7-peak spectral modeling³. Slice-by-slice subtraction of the pre and post-contrast DIXON images generated conventional DCE MRA images, while water-only images from the post-contrast mDIXON were used as the subtractionless images. Maximum-intensity projections (MIPs) of both sets of images, subtraction and water–only, were generated. Vessel-to-background contrast of selected arterial segments (Figure 1, Table) were evaluated by two radiologists, who also rated, in a blinded manner, the overall image quality from 0-3 (0=non-diagnostic, 3=good quality without visible artefacts). In addition, quantitative measures of vessel-to-background contrast of the same segments were performed with region of interest (ROI) analysis (see the ROIs on Fig1C). Paired, one-sided t-test was used to test for significant differences (p<0.05) in the vessel-to-background contrast of the two techniques.

Figure 1 shows a comparison of a standard subtraction (1A) vs. ‘subtraction-less’ (1B) MRA of the neck from the same patient. While high vessel-to-background contrast was observed in both methods, the ‘subtraction-less’ DIXON-MRA resulted in higher overall vessel-to-background ratio for the aortic arch (5.6%), brachiocephalic arteries (7.0%), right carotid bifurcation (23.2%), and left carotid bifurcation (20.4%) (see Table). The ROIs used in the evaluation are shown on Fig. 1C (different patient), which also shows the recovery of the missing signal (1B, white arrowhead) in the distal subclavian arteries when the shim was positioned over the chest (yellow rectangle).The petrous portions of the carotid arteries did not show any significant difference in vessel-to-background ratio, which may be due to the slight increase in the brain parenchymal background signal observed when no subtraction was used (white arrow, Fig 1B).Dixon-based ‘subtraction-less’ MRA of the neck allows for capturing high-spatial resolution DCE images of the aortic arch, the subclavian arteries, and the carotid bifurcation over extended FOVs and with improved vessel-to-background contrast compared to subtraction MRA.