- Convolutional Neural Network for Automated FLAIR Lesion Segmentation on Clinical Brain MR Imaging
This convolutional neural network was retrospectively trained on 295 brain MRIs to perform automated FLAIR lesion segmentation. Performance was evaluated on 92 validation cases using Dice scores and voxelwise sensitivity and specificity, compared with radiologists' manual segmentations. The authors' model demonstrated accurate FLAIR lesion segmentation performance (median Dice score, 0.79) on the validation dataset across a large range of lesion characteristics. Across 19 neurologic diseases, performance was significantly higher than existing methods (Dice, 0.56 and 0.41) and approached human performance (Dice, 0.81).
- The Interpeduncular Angle: A Practical and Objective Marker for the Detection and Diagnosis of Intracranial Hypotension on Brain MRI
MRIs of 30 patients with intracranial hypotension and 30 age-matched controls were evaluated by 2 neuroradiologists for classic findings of intracranial hypotension and the interpeduncular angle. Group analysis was performed with a Student t test, and receiver operating characteristic analysis was used to identify an ideal angle threshold to maximize sensitivity and specificity. The interpeduncular angle had excellent interobserver reliability (intraclass correlation coefficient value = 0.833) and was significantly lower in the intracranial hypotension group compared with the control group (25.3° versus 56.3°). There was significant correlation between the interpeduncular angle and the presence of brain stem slumping. With a threshold of 40.5°, sensitivity and specificity were 80% and 96.7%, respectively.