Dynamic 3D MR Angiography of Intra- and Extracranial Vascular Malformations at 3T: A Technical Note

Description of Techniques

All MR imaging measurements were performed on a 3T whole-body system by using an eight-element phased-array head coil. Dynamic MRA was performed by using a 3D fast low-angle shot (FLASH) sequence (12) with the following parameters: TE/TR, 1.62/4.31 ms; bandwidth, 490 Hz/Px; partially sampling a k-space volume of 256 × 256 × 16 data points, with an in-plane image resolution of 0.8 × 0.8 mm and a section resolution of 6 mm. Parallel imaging with a GRAPPA factor (generalized autocalibrating partially parallel acquisitions) of two was applied in phase-encoding direction sampling of 24 reference lines in the k-space center (13). This technique allows the acquisition of one 3D data set with the described sequence parameters in 1.5 seconds. The measurements were repeated 22 (dAVF and eAVM) and 25 times (iAVM). Contrast injection was started 7 seconds after start of measurement. To eliminate background nonvascular signal intensity, a mean mask from the first four data sets was calculated and image subtraction of the dynamic data sets from the precontrast mask was performed. Further processing and image evaluation was then done on the subtraction images. A standard dose of 0.1 mmol/kg body weight gadolinium-gandobenate dimeglumine (Multihance, Altana, Konstanz, Germany) followed by 20 mL of physiologic saline solution was applied via the cubital vein. Because no injection system for a 3T unit was commercially available at this time, a manual injection of about 2 mL/s was performed.

Applications

Discussion

Dynamic 2D MR projection angiography was primarily described by Wang et al (1) and modified by Hennig et al (2), reaching a temporal resolution in the subsecond domain. One of the major drawbacks of these time-resolved methods is the low signal intensity-to-noise ratio due to the fast acquisition time. Processing strategies derived from functional imaging as correlation analysis can be applied, leading to an increase of the signal intensity-to-noise ratio and a reduction of background noise (3). Cerebral dural arteriovenous fistulas (5–7), intracranial vascular malformations (4, 8), and extracranial vascular malformations (9) have been investigated by using contrast-enhanced 2D projection techniques. Meanwhile, time-resolved techniques give a better temporal resolution by using modified methods of k-space filling, and 3D acquisitions can be obtained at a higher spatial resolution. Coley et al (14) reached a temporal resolution of one frame per second by using a radial-projection sliding window technique. Time-resolved contrast-enhanced 3D MRA of the whole brain could be performed at a temporal resolution of one frame per 1.8–2.1 second by using parallel imaging at 1.5T (15).

In this technical note, dynamic contrast-enhanced 3D MRA at 3T combined with parallel imaging was performed by taking advantage of the reduction of acquisition time established by parallel imaging and the higher signal intensity-to-noise ratio at 3T. A temporal resolution of one 3D data set every 1.5 seconds could be achieved by using a parallel imaging technique. The expected improvement of the signal intensity at 3T in combination with the saturation effects onto the stationary background tissue (due to the prolonged T1 relaxation times at 3T) was additionally advantageous.

The temporal resolution was sufficient for the extracranial and intracranial AVM. In the high-flow iAVM an arteriovenous transit time of 1 second could be detected. A normal transit time of 3 seconds in this case could be established in healthy parts of the vasculature and is confirmed by the findings of Noguchi et al (7). For the high-flow dAVF, the arteriovenous transit time was so fast that a simultaneous and identical time course of affluxion artery and draining vein was found. But this finding could be confirmed by conventional angiography showing a simultaneous appearance of arterial feeder and drainage vein.

The in-plane spatial resolution was only slightly improved from 0.97 × 0.97 mm² at the 2D MRA (5, 9) to 0.8 × 0.8 mm² at the herein presented 3D MRA to increase temporal resolution. The section thickness, however, was different: 80 mm for the 2D MRA and 6 mm for the 3D MRA (with 16 sections acquired).

The dominant feeding arteries were sufficiently depicted in the iAVM and the eAVM. The small terminal branch of the affluxion artery of the dAVF could only be suspected. The low diameter of the vessel and the resulting low signal intensity from this feeding artery are accounted as reasons for the missing depiction. The nidus of the eAVM could be clearly depicted, whereas the nidus of the iAVM did not reach sufficient signal intensity to be clearly assessed.

Conclusion

Dynamic 3D MRA combined with parallel imaging could be shown to be advantageous by receiving at temporal resolution of one image per 1.5 seconds which is fast for 3D MRA. Further improvement of temporal resolution may derive from stratified versions of k-space filling into the subsecond domain. For the intracranial and extracranial AVM, the temporal resolution was sufficient by even separating pathologically and normally filled intracranial veins. For the high-flow dural arteriovenous fistula, however, a simultaneous filling of arterial feeder and venous drainage was found by dynamic 3D MRA and conventional angiography.