Diagnosis of spinal dural arteriovenous fistula using 3D T2-weighted imaging.

Abstract

To evaluate spinal MRIs without and with 3D T2W imaging among patients without and with spinal dural arteriovenous fistula (SDAVF) confirmed by spinal digital subtraction angiography (DSA). A retrospective case-control study was performed among patients without and with SDAVF who had both spinal MRIs and gold standard spinal DSA. Two neuroradiologists independently reviewed spinal MRIs that were performed with either sagittal T2W turbo spin echo (2D group) or sagittal 3D T2W sampling perfection with application-optimized contrasts using different flip-angle evolutions (SPACE) (3D group) and documented the presence or absence of SDAVF. Using spinal DSA diagnosis as a gold standard, the sensitivity, specificity, and interobserver agreement for the 2D-group and 3D-group MRI diagnosis were calculated. The 2D group consisted of 21 patients and the 3D group consisted of 16 patients. For both radiologists, the 2D group demonstrated a sensitivity of 100% and specificity of 100%. Interobserver agreement in the 2D group was perfect (k=1.0). For both radiologists, the 3D group demonstrated sensitivity of 100.0% and specificity of 92.3%. Interobserver agreement in the 3D group was perfect (k=1.0). While flow voids were considered more conspicuous, spinal cord signal abnormality was considered less conspicuous with 3D T2W SPACE compared with conventional 2D STIR sequence. 3D T2W SPACE should be used in conjunction with 2D T2W sequences to more accurately detect abnormal cord signal and determine when perimedullary flow voids are pathologically abnormal for the radiologic diagnosis of SDAVF.