Identification of the feeding arteries of spinal vascular lesions via phase-contrast MR angiography with three-dimensional acquisition and phase display.

PURPOSE: To determine whether identification of the feeding arteries of spinal vascular lesions with phase-contrast MR angiography benefits from the higher spatial resolution of three-dimensional (volume) acquisitions and flow-direction information provided by the phase reconstruction of two-dimensional acquisitions. METHODS: Fifteen patients with high- or low-flow spinal vascular lesions proved by spinal arteriography underwent MR angiography with phase-contrast techniques. Arteriographic and MR angiographic studies were reviewed to identify the arterial feeders of spinal vascular lesions. RESULTS: On modulus reconstructions of coronal 2-D or 3-D acquisitions, three of four arteries feeding high-flow lesions and three of 14 arteries feeding low-flow lesions were identified as hypertrophic vessels joining the parent intercostal or cervical arteries. Of 11 intradural veins draining dural arteriovenous fistulas, three were identified on coronal 2-D acquisitions and six on coronal 3-D acquisitions as vessels that coursed from a neural foramen to a midline tangle of vessels. Phase reconstruction showed ascending and descending flow patterns in two patients with intramedullary arteriovenous malformations, and diverging flow in perimedullary veins draining a hemangioblastoma. In nine patients with dural arteriovenous fistulas, phase reconstruction provided information as to the level of the arterial feeders. Phase reconstruction in coronal plane acquisitions also provided evidence of centripetal flow. CONCLUSION: Three-dimensional acquisitions and phase display of 2-D acquisitions improved the visibility of arterial pedicles of spinal vascular lesions at phase-contrast MR angiography.

MRI of spinal epidural lymphoma.

We reviewed the MRI features in eight patients with spinal epidural lymphoma (clinically primary in 4 patients); one patient had multiple lesions. The cervical spine was involved in one patient, the thoracolumbar spine in 5 and the sacrum in two. Mean longitudinal extension of the epidural lesion was 2.6 vertebral segments. The tumours were homogeneously isointense with the spinal cord on T1-weighted images and isointense or hyperintense on proton-density and T2-weighted images. The spinal cord was compressed in four patients but showed signal changes in only one. In five patients the lesions communicated through the intervertebral foramina with paravertebral soft tissue masses. In all but one of the patients diffuse signal changes in the vertebral body marrow consistent with osteolytic or osteoblastic changes were identified adjacent to or at distance from the epidural lesion. Vertebral collapse was observed in two patients.