Sequential three-dimensional time-of-flight MR angiography of the carotid arteries: value of variable excitation and postprocessing in reducing venetian blind artifact.

OBJECTIVE

Multiple overlapping three-dimensional (3D) time-of-flight carotid MR angiography potentially combines many of the desirable features of two-dimensional (2D) and single-volume 3D MR angiographic imaging techniques. Yet the maximum-intensity-projection images from such acquisitions are often degraded by artifact due to nonuniform signal intensity of contiguous imaging volumes and inadequate, yet arduous, postprocessing. The former has been termed venetian blind artifact. To date, the severity of the artifact has been minimized by the use of very thin slabs with a large percentage of overlap. However, the artifact typically is still appreciable, and the required acquisition and postprocessing times are increased. The purpose of this study was to examine the value of technical modifications of both the multislab acquisition and postprocessing procedures to reduce this artifact on images of healthy volunteers.

SUBJECTS AND METHODS

Spatially variable RF pulses along the direction of flow were applied as excitation pulses in the multislab time-of-flight MR angiographic acquisitions to compensate for the nonuniform blood signal intensity caused by spin saturation. An automatic postprocessing technique was used to optimally combine the image data in overlapping slices by selecting the higher-intensity pixel of the two on a pixel-by-pixel basis. Ratios of the standard deviation of signal intensity to the mean signal intensity were computed as a function of RF profile and postprocessing method along the long axes of the arteries to measure the uniformity of the signal intensity of the blood. The spatially variable and sinc RF pulse acquisitions, combined with automatic and conventional manual postprocessing, were compared.

RESULTS

Compared with the sinc pulse acquisition, the MR angiograms acquired with spatially variable excitation pulses improved the signal uniformity of the arteries with thicker volumes and less overlap, thereby reducing the acquisition time by 25% for similar spatial coverage. When used with the automatic postprocessing technique, the severity of the venetian blind artifact on maximum-intensity-projection images was minimized and the postprocessing time was reduced by roughly a factor of 5.

CONCLUSION

The combined use of spatially variable excitation pulses and an automatic postprocessing technique can improve the uniformity of the signal from blood across the slab and allow thicker slabs to be acquired with less overlap. Data acquisition and postprocessing times can be reduced significantly. This work suggests it may be possible to easily produce overlapping 3D MR angiograms that should be superior to conventional 2D and 3D studies.