Maximizing the Conspicuity of CSF-Venous Fistulas on CT Myelography: Assessment of Contrast Density and Timing Effects.

BACKGROUND AND PURPOSE: Advancements in CT myelography (CTM) have improved visualization of CSF-venous fistulas (CVFs), a frequent cause of spontaneous intracranial hypotension (SIH). However, the relative impact of the timing of image acquisition and the contrast density in the subarachnoid space remain unclear. This study compared the effects of timing and contrast density in the ipsilateral subarachnoid space and assessed the impact of other technical factors on CVF conspicuity using a validated instrument to stratify diagnostic confidence. MATERIALS AND METHODS: A retrospective review of International Classification of Headache Disorders, 3rd edition-confirmed patients with SIH with CVFs was performed. Only fistulas classified as definite by the Duke CSF-Venous Fistula Confidence Score (DCCS) were included. All available CTMs covering each index fistula site were reviewed, excluding examinations occurring after surgery or embolization for a definite CVF. We assigned a DCCS to each acquisition and recorded contrast density in the subarachnoid space ipsilateral to the known CVF and image acquisition time. Patient positioning and scanner type were also collected as potential confounders. Ordinal logistic regression was used to assess associations with CVF conspicuity. RESULTS: One hundred forty-four patients with 149 definite CVFs comprised the final cohort, from which 222 CTMs and 697 acquisitions were assessed. Both increased contrast density and reduced acquisition time were associated with increased CVF conspicuity in univariate analyses (P < .001). When adjusting for sex, scanner type, and patient positioning in the multivariate model, contrast density and time remained significant predictors of conspicuity (P < .001). Density had a 4-fold greater impact on conspicuity than time, with a 14.3% increase in likelihood of CVF detection per 100 Hounsfield unit (HU) increase in attenuation and an optimal target threshold at 836 HU. CONCLUSIONS: Both contrast density and time influence conspicuity of CVFs on CTM; however, the greater relative impact of density suggests that myelogram technique should prioritize maximization of contrast density for optimal visualization of CVFs. Positioning strategies to increase local contrast pooling may improve CVF detection more effectively than timing adjustments alone.

Duke Scholars

Author Josh Lim  

Author Peter George Kranz Radiology, Neuroradiology

Author Timothy James Amrhein Radiology, Neuroradiology