| Abstract |
Background: Intracranial vessel tortuosity is a key component of dolichoectasia and has been associated with atherosclerosis and adverse neurologic outcomes. However, the evaluation of tortuosity is mainly a descriptive assessment.
Purpose: To compare the performance of three automated tortuosity metrics (angle metric [AM], distance metric [DM], and distance-to-axis metric [DTA]) for detection of dolichoectasia and presence of segment-specific plaques.
Study type: Observational, cross-sectional metric assessment.
Population: 1899 adults from the general population; mean age = 76 years, female = 59%, and black = 29%.
Field strength/sequence: 3-T, three-dimensional (3D) time-of-flight MRA and 3D vessel wall MRI.
Assessment: Tortuosity metrics and mean luminal area were quantified for designated segments of the internal carotid artery, middle cerebral artery, anterior cerebral artery, posterior cerebral artery, vertebral artery, and entire length of basilar artery (BA). Qualitative interpretations of BA dolichoectasia were assessed based on Smoker's visual criteria.
Statistical tests: Descriptive statistics (2-sample t-tests, Pearson chi-square tests) for group comparisons. Receiver operating characteristics area under the curve (AUC) for detection of BA dolichoectasia or segment-specific plaque. Model inputs included 1) tortuosity metrics, 2) mean luminal area, and 3) demographics (age, race, and sex).
Results: Qualitative dolichoectasia was identified in 336 (18%) participants, and atherosclerotic plaques were detected in 192 (10%) participants. AM-, DM-, and DTA-calculated tortuosity were good individual discriminators of basilar dolichoectasia (AUCs: 0.76, 0.74, and 0.75, respectively), with model performance improving with the mean lumen area: (AUCs: 0.88, 0.87, and 0.87, respectively). Combined characteristics (tortuosity and mean luminal area) identified plaques with better performance in the anterior (AUCs ranging from 0.66 to 0.78) than posterior (AUCs ranging from 0.54 to 0.65) circulation, with all models improving by the addition of demographics (AUCs ranging from 0.62 to 0.84).
Data conclusion: Quantitative vessel tortuosity metrics yield good diagnostic accuracy for the detection of dolichoectasia.
Level of evidence: 1 TECHNICAL EFFICACY STAGE: 2.
|
