Magnetic resonance angiography (MRA) methods such as Time of Flight (ToF) can benefit from fat suppression techniques in areas where the fat signal obscures the contrast between areas of flow and static tissue [1]. The three-point Dixon water-fat separation methods have been shown to provide a means of removing fat signal with a high degree of robustness to B0 field inhomogeneity over conventional spectral saturation pulse techniques [2–5]. This work shows how the Dixon method can also improve upon the “out-of-phase” technique where the echo time is chosen so that fat is out of phase in the acquired signal [6]. The proposed method also uses the spiral k-space trajectory in [7] and its combination with the Dixon method as detailed in [2] for application to time of flight (ToF) angiography in the head and neck. The spiral acquisition technique improves the scan efficiency allowing the acquisition of the three data points necessary for Dixon reconstruction while improving upon the scan time of the conventional 3D multi-slab cartesian ToF sequence.

Scan Parameters: The proposed method was compared to conventional cartesian methods for 3D multi-slab imaging of the circle of Willis and 2D multi-slice imaging of the carotid bifurcation. All scans were performed on a volunteer using a 3T Philips Ingenia system with a 15 channel head coil.

3D ToF: The standard cartesian parameters used were: TR/TE = 23/3.5 msec, 16 x 16 cm field of view (FOV), 18˚ flip angle, a sampling window (tau) of 5.7 msec and a partial echo acquisition of 65% in the readout direction for a scan time of 5:25 min. The proposed spiral scan parameters were: TR = 21 msec, TEs = 1.24/1.99/2.74 msec, 24 x 24 cm FOV, 18˚ flip angle, tau = 7.2 msec, and 40 spiral interleaves per kz-encoding (fully sampled) for a scan time of 4:02 min. Both sequences had a 1.4 mm slice thickness, 19 slices per slab with an overlap of 9 slices for 5 slabs, flow compensation gradients and a venous saturation pulse.

2D ToF: The cartesian scan parameters used were: TR/TE = 16/3.5 msec, 22 x 22 cm FOV, 30˚ flip angle, tau = 3.6 msec, and a partial echo of 65%, for a scan time of 4:46 min. The proposed spiral scan parameters were: TR = 25 msec, TEs = 2.5/3.25/4 msec, 35˚ flip angle, tau = 8.8 msec, and 28 spiral interleaves per slice (fully sampled) for a scan time of 3:34 min. The parameters used in both techniques were: 3 mm slice thickness with an overlap of 1 mm, flow compensation and a venous saturation pulse.

Reconstruction: The reconstruction for the proposed 2D and 3D spiral methods was implemented and processed in GPI [8]. In each reconstruction, the three point Dixon method was used to calculate the B0 field maps. The field maps are then used to reconstruct the three echo sets in a joint conjugate gradient deblurring and water-fat separation [9].

The thin maximum intensity projection (MIP) images in figure 1 demonstrate that the removal of the fat signal allows for better visualization of the ophthalmic arteries, a region that can be significantly obfuscated by fat signal. Figure 2 shows how the fat signal in the orbitals and surrounding scalp regions can come through in the full volume MIP images of the head. MIP images of the carotid bifurcation (figure 3) show that fat signal is removed from areas around the neck and posterior head. This work has demonstrated the viability of using spiral and the three-point Dixon method for water-fat separation in ToF angiography with a total scan duration that is shorter than conventional methods. The authors believe that the scan duration can be made substantially shorter by increasing the sampling window duration as shown in [10].