Study of magnetic resonance imaging-based arteriovenous malformation delineation without conventional angiography

In this study, we aimed to assess the feasibility of arteriovenous malformation (AVM) delineation for gamma knife radiosurgery without conventional angiography and to correlate factors that may affect AVM delineation.

A series of 57 consecutive patients with AVMs treated with gamma knife radiosurgery from August 1994 to December 2000 were reviewed. All patients in the study had undergone pretreatment angiography. The mean AVM volume was 2.8 cm(3), with a median of 2.0 cm(3) (range, 0.04-22 cm(3)). All AVMs were delineated on the original frame-based magnetic resonance imaging (MRI) scans by a vascular neurosurgeon without the assistance of angiography and then compared with the actual AVM delineation on the basis of previously performed angiography and MRI. Univariate correlation analysis was used to determine the relationship of AVM coverage, size, diffuseness, previous embolization, and hemorrhage parameters.

The study volume or MRI-based volume alone coincided with the actual treatment volume by a mean of 58% for diffuse and 87% for nondiffuse AVMs (P = 0.0005). At AVM volume greater than 2 cm(3), the median percentage of coinciding volume was 63% for embolized AVMs and 82% for nonembolized AVMs (P = 0.0315). Conversely, the study volume overestimated the actual treatment volume by a mean of 57% for AVMs larger than 2 cm(3) versus 25% for AVMs smaller than 2 cm(3) (P = 0.0012). In general, the percentage of the coinciding volume was inversely related to that of the excess volume, whereas both the study volume and the coinciding volume were proportionate to AVM volume at treatment.

MRI-based AVM delineation without conventional angiography may be feasible only for selected patients, such as those with nondiffuse and large nonembolized AVMs.