Sinus- and Sequence-Specific Diagnostic Performance of Routine Unenhanced Brain MRI in Dural Venous Sinus Thrombosis

Background: Dural venous sinus thrombosis (DVST) is an uncommon but potentially life-threatening cerebrovascular disorder requiring early diagnosis to prevent serious complications. Although CE-MRV is the reference standard, routine brain MRI is often the first imaging study in patients with nonspecific neurological symptoms, and the sinus-specific diagnostic performance of individual sequences remains incompletely defined. Purpose: To evaluate the diagnostic performance and inter-reader agreement of routine brain MRI sequences for DVST detection using a sinus-specific framework. Methods: This retrospective case-control study included 140 patients (34 with DVST, 106 age-matched controls) imaged on 1.5 T and 3.0 T scanners. Two blinded neuroradiologists evaluated six unenhanced sequences (sagittal/axial T1WI, T2WI, FLAIR, DWI [b = 1000 s/mm²], and SWI) across four dural sinuses, using CE-MRV and CE-3D T1WI as the reference standards. Logistic regression and Cohen's κ assessed diagnostic performance and inter-reader agreement, respectively. Results: Globally, DWI with FLAIR achieved 97.9% accuracy, 91.2% sensitivity, and 100% specificity (AUC = 0.997). Optimal sequences varied by sinus: sagittal T1WI with SWI for the superior sagittal sinus (accuracy = 99.3%), DWI with SWI for the transverse sinus (97.9%), DWI with FLAIR and T2WI for the sigmoid sinus (98.6%), and SWI with axial T1 for the straight sinus (100%). Inter-reader agreement was substantial to almost perfect for routine sequences (mean κ = 0.874) and almost perfect for CE-MRV and CE-3D T1WI (κ = 0.98). Conclusions: Routine brain MRI provides reliable DVST detection with a sinus-tailored multisequence strategy. DWI and FLAIR offer robust diagnostic performance in global evaluation, while T1WI, SWI and T2WI add segment-specific value, reserving CE-MRV and CE-3D T1WI for equivocal or clinically suspicious cases.