Nederkoorn et al (1) conclude that flow voids on nonenhanced 3D time-of-flight (TOF) MR angiography (MRA) images represent severe carotid artery stenosis. Although they present compelling evidence that this conclusion is correct for their particular MR imaging system with their particular imaging parameters, radiologists should be advised of the peril of generalizing these results to any MR imaging system, using any imaging parameters. Specifically, preliminary data from our neurovascular lab suggest that the presence of flow voids on 2D TOF MRA images, for a given degree of carotid artery narrowing, is critically dependent on choice of echo time (TE) for the TOF pulse sequence, specific MR imaging hardware, or both.
In a pilot study of patients who underwent both carotid duplex sonography and 2D TOF MRA for evaluation of suspected internal carotid artery stenosis, 20 were imaged on a newer LX unit (GE Medical Systems, Milwaukee, WI) by using a short-TE pulse sequence (TE ≈4.7 ms), and 24 were imaged on an older Signa unit (GE Medical Systems) by using a long-TE pulse sequence (TE ≈8.7 ms). Of the 20 imaged with the short-TE pulse sequence, TOF signal dropout was seen in one (100%) of one with hairline lumen, in three (50%) of six with peak systolic velocity (PSV) more than 400 cm/s, in four (50%) of eight with PSV between 200 and 400 cm/s, and in none (0.0%) of three with PSV less than 200 cm/s (two patients with PSV’s of ∼370 and 540 cm/s had equivocal signal dropout). Of the 24 imaged with the long-TE pulse sequence, TOF signal dropout was seen in one (100%) of one with hairline lumen, in 10(100%) of 10 with PSV more than 400 cm/s, in four (80%) of five with PSV between 200 and 400 cm/s, and in one (14.3%) of seven with a PSV less than 200 (one patient with PSV ∼300 cm/s had equivocal signal dropout). One patient was imaged twice, each imaging session a week apart without interval treatment, by using different TE values. The first images, which were obtained with a long TE of 8.6 ms, showed a flow void, whereas the follow-up images, which were obtained with a short TE of 4.7 ms, did not.
These findings are consistent with the fact that flow voids on TOF MRA images are caused by intravoxel dephasing and are thus less likely to occur with short than with long TEs. Additionally, the stronger gradients and more homogeneous magnetic fields present in newer MR units, which permit smaller voxel sizes, may also predispose to decreased intravoxel dephasing, and hence lower sensitivity for signal dropout from turbulent flow. Although, as Nederkoorn et al point out, 3D TOF MRA techniques “have higher spatial resolution, a greater signal-to-noise ratio, and lower sensitivity for voids because of the smaller voxels and shorter echo time(s)” as compared with those of 2D TOF MRA, the precise relationship between 3D TOF flow void detection thresholds and the specific MR imaging hardware and software used has yet to be determined. Until it has, we continue to advise a conservative approach to flow void interpretation on TOF MRA images. Indeed, radiologists ideally should calibrate TOF signal dropout for their particular MR units and pulse sequences with an external reference standard of stenosis, such as sonography or CTA, before image interpretation. This may be especially prudent in some clinical environments wherein surgeons consider patients with carotid artery flow voids to have “proved” severe (>70%) lumenal stenosis, and therefore to be candidates for carotid endarterectomy.
References
- Copyright © American Society of Neuroradiology
