Diagnostic Performance of MRI Sequences for Evaluation of Dural Venous Sinus Thrombosis

OBJECTIVE: The purpose of this study is to assess the performance of routinely used MRI sequences with and without contrast enhancement in the diagnostic evaluation of dural venous sinus thrombosis (DVST). MATERIALS AND METHODS: We identified consecutive patients older than 18 years who underwent concurrent standardized brain MRI and contrast-enhanced (CE) MR venography (MRV) examinations for suspected DVST. The seven MRI sequences that were used (axial unenhanced T1-weighted, T1-weighted CE, T2-weighted, DWI, T2-weighted FLAIR, T2-weighted gradient-recalled echo [GRE], and sagittal 3D T1-weighted GRE CE sequences) were randomized, anonymized, and reviewed independently by two neuroradiologists who were blinded to the final diagnosis. Ten separate venous sinus segments were evaluated. CE MRV was the reference standard for determining the presence or absence of DVST, and it was performed using the following imaging parameters: TR/TE, 4.1-77/1.4-9.5; flip angle, 12-35°; and slice thickness, 0.8-1.4 mm. The diagnostic performance of and interobserver variability for each sequence was assessed per patient and per segment. RESULTS: Thirty-six patients with DVST (72% of whom had acute thrombosis and 28% of whom had chronic thrombosis) and 29 patients without DVST were included in the study. For each sequence, the AUC values for the detection of DVST per patient, as determined by reviewer 1 and reviewer 2, respectively, were as follows: for T1-weighted unenhanced sequences, 55% and 61%; for T1-weighted CE sequences, 79% and 80%; for T2-weighted sequences, 77% and 76%; for DWI sequences, 59% and 64%; for T2-weighted FLAIR sequences, 70% and 72%; for T2-weighted GRE sequences, 64% and 66%; and for the 3D T1-weighted GRE CE sequence, 77% and 81%. The diagnostic performance of the 3D T1-weighted GRE CE sequences was statistically significantly greater than that of the other sequences. Interobserver variability ranged from 0.26 (for T1-weighted unenhanced sequences) to 0.73 (for the DWI sequence). Overall, for each reviewer and with the use of all evaluated sequences, MRI had a high sensitivity (> 99% for both reviewers) but low specificity (14% for reviewer 1 and 48% for reviewer 2) for the detection of DVST. CONCLUSION: Sequences used in routine brain MRI performed with and without contrast enhancement have varying strengths that are important to recognize when the likelihood of DVST is assessed, but they do not replace the utility of dedicated CE MRV.