Radiologic Findings of Deep Seated Cere b ral Arteriovenous

PURPOSE: To analyze the radiologic-especially angiographic-findings of deep seated cerebral arteriovenous malformation(AVM) involving nonvisualized straight sinus. MATERIALS AND METHODS: In six patients aged between 15 and 53 years with deep seated cerebral AVM, CT and MR images were retrospectively analyzed with regard to the following features : the presence of straight sinus, the location of AVM, and the occurrence of hemorrhage. Angiograms were analyzed for venous drainage routes of AVM, the appearance of veins, the presence of falcine sinus and venous drainage from normal deep brain parenchyme. In four patients who had undergone intravascular embolization therapy, pre- and post- embolization angiograms were compared. RESULTS: CT and MR images showed neither straight sinus nor thrombosis. AVMs were deeply seated in the brain, and in all cases there was cerebral hemorrhage. Angiograms disclosed that venous drainage of all AVMs occurred via the veins of Galen. In one case, venous flow via the falcine sinus to the superior sagittal sinus was noted, but in others, retrograde flow in the deep venous system was observed. Marked collateral routes followed in response to the obstruction of straight sinus included the basal vein of Rosenthal, the internal occipital, internal cerebral, and cerebellar hemispheric veins (which are Galenic afferents), and the inferior sagittal s-inus. In all patients, contralateral routes were partially involved. Venous drainage from normal deep parenchyme through the transcerebral veins to the superficial venous system was noted, and in one case, s-traight sinus which had been observed on an angiogram five years earlier was no longer present. CONCLUSION: Angiography offers effective evaluation of the dynamic aspect of venous flow in cases involving deep-seated AVM, and of normal deep parenchyme in cases in which AVM involves nonvisualized straight sinus. Before intravascular treatment of AVM, venous flow must be carefully analyzed.