Automatic identification of Moyamoya disease based on time-of-flight MR angiography

Moyamoya disease (MMD) is a rare chronic progressive cerebrovascular disease that causes strokes. For the diagnosis of MMD, time-of-flight magnetic resonance angiography (TOF-MRA) can be an alternative to digital subtraction angiography (gold standard) owing to its non-invasive and radiation-free attributes. In this study, the deep learning method (ResNet-50) was used for MMD automatic diagnosis on the maximum intensity projection images from 3D TOF-MRA, and five-fold cross-validation was used for validation. The method exhibits the accurate ability (AUC: 0.990 ± 0.008, accuracy: 0.933 ± 0.063) to identify MMD and has the potential to improve the clinical management of MMD.