Intracranial hemorrhage is a leading cause of death and disability in patients with moyamoya disease (MMD), yet clinically available tools for predicting long-term hemorrhage risk are lacking. Dilatation and rupture of fragile collateral vessels are considered the main cause of MMD-related hemorrhage; however, conventional 1.5T/3.0T MRI cannot adequately quantify collateral vessel morphology, blood flow, and vessel wall features. Metabolomic alterations have been implicated in MMD pathogenesis, including angiogenesis and collateral vessel formation.
This prospective cohort study will consecutively enroll patients with MMD or moyamoya syndrome confirmed by digital subtraction angiography (DSA) at Beijing Hospital. All participants will undergo preoperative 5.0T brain MRI, including time-of-flight MR angiography (TOF-MRA), 4D MRA, high-resolution vessel wall imaging (HR-VWI), 3D arterial spin labeling (ASL), and conventional sequences (T1WI, DWI, T2WI, SWI). Preoperative blood and urine samples will be collected for metabolomic profiling, and superficial temporal artery (STA) specimens trimmed during direct bypass surgery will be retained for immunohistochemical cross-validation.
Using deep learning (VT-UNet for collateral vessel segmentation and Swin Transformer for feature extraction) combined with machine learning methods, the investigators will develop and validate an integrated hemorrhage risk prediction model that combines 5T MRI features, hemodynamic parameters, metabolomic biomarkers, and clinical baseline data. Participants will be followed for 12 months; new-onset intracranial hemorrhage, analyzed per cerebral hemisphere, will be the endpoint event.
The study aims to (1) establish a hemorrhage risk stratification system for early identification of high-risk patients to guide timely surgical revascularization, and (2) identify core metabolomic and imaging biomarkers of MMD-related hemorrhage.