Time-resolved, non-contrast-enhanced, MR angiography (NCE-MRA) is an attractive alternative to x-ray digital subtraction angiography (DSA) or contrast-enhanced MRA (CE-MRA), by providing dynamic blood flow patterns with high spatial and temporal resolution. Current dynamic NCE-MRA techniques^1-3 are all developed based on the arterial spin labeling (ASL), which require twice of the imaging time for obtaining control/label pairs and are prone to subtraction errors from motion-induced misregistration. The recently introduced Fourier-Transform based velocity-selective saturation (FTVSS) pulse trains^4-6 (setting flowing spins in the pass-band and static tissues in the saturation-band) can preserve the angiographic signal and suppress tissue background within a single scan (VSMRA). Thus a tissue mask can be generated by thresholding the signal of obtained raw data. Here we demonstrate the time-resolved cerebral MRA at 3T using FTVSS pulse trains in the last phase and a VSMRA-derived tissue mask to remove stationary tissue background.

The pulse sequence diagram for this inflow-based dynamic MRA is shown in Fig. 1. The cycle starts with a slab-selective saturation pulse to suppress all the spins in a volume including both the imaging plane and the downstream area, followed by series of segmented acquisition modules with constant phase intervals. The last phase is preceded by a repetition of two consecutive FTVSS pulse trains to highlight the vascular signal while suppressing the tissue background (VSMRA). Each FTVSS pulse train is 48ms and contains a series of 9 excitation pulses (10° each), interleaved with pairs of 180° refocusing pulses each of which is surrounded by a pair of encoding gradient lobes with alternating polarity⁵ (Fig. 1). The velocity field of view was set to be 45cm/s with a targeted saturation band within ±4cm/s.

Experiments were performed on a 3T Philips Achieva scanner using the body coil for transmission and a 32-channel head coil for reception. A total of 6 healthy volunteers (age: 25~51yrs) were enrolled with informed consent. For scan planning purposes, phase contrast MRA (PCA) images were acquired in both the coronal and sagittal planes. Time-resolved MRA were tested for 1) a coronal orientation covering both the major extracranial and intracranial vessels (Fig. 2a-c) and 2) an axial plane through the circle of Willis (Fig. 2d-f). Both imaging slabs are 70mm thickness with 1mm isotropic obtained with 3D turbo field echo (TFE) acquisition: TR/TE=10/2ms, flip angle=11°, TFE factor=10, TFE acquisition window=100 ms, SENSE 3x1. A total of 10 phases were sampled with a phase interval of 110ms for the first 9 phases and repetition time of each cycle is 1600ms. Durations of coronal and axial scans were 7.3min and 5.6min, respectively. In post processing, an empirical threshold was set for the VS-MRA data from the last phase to generate a binary mask (tissue: 0; blood: 1). Compared to the tissue in the brain, surrounding fat, skull and muscle at peripheral regions were less suppressed due to B0/B1 inhomogeneities and were thus manually extracted from the raw images. Multiplying this tissue mask to all other data reveals dynamic MRA through different phases after maximum-intensity-projection (MIP).

Fig. 3a displays representative MIP images from the multi-phase raw data of a coronal scan. For the first 9 phases tissue signal is gradually recovered from the initial saturation and only large vessels such as carotid arteries are visible; the FTVS prepared MRA at the10th phase yields significant suppression of tissue background, as expected. Fig. 3b exhibits the final images after applying the tissue mask obtained from the VSMRA, showing dynamic filling of inflow arterial blood from internal carotid arteries (ICA) and vertebral arteries (VA) to circle of Willis and then to middle cerebral arteries (MCA), anterior cerebral arteries (ACA), and posterior cerebral arteries (PCA). Similar results for the axial scans are arrayed in Fig. 4a and 4b, with a clear depiction of arterial blood flow from large segments of MCA, ACA and PCA to their small distal branches.A time-resolved NCE-MRA technique has been developed at 3T with background suppresstion by applying a tissue mask, which is produced from thresholding the VSMRA obtained in the same phase cycle. The technical feasibility was shown by the excellent depiction of the dynamic blood flow of cerebral arteries in healthy subjects. Our next step is to compare the diagnosis in patients with cerebral vascular disorders by the proposed method with that by the established methods such as CE-MRA.