Hypoperfusion Intensity Ratio as a Predictive Marker in Pre-Endovascular Treatment Perfusion MRI for Acute Middle Cerebral Artery Occlusion with Intracranial Atherosclerosis

Introduction Identifying the cause of MCA occlusion before endovascular treatment (EVT) in acute ischemic stroke is useful. Hypoperfusion intensity ratio (HIR) is defined as the volumetric ratio of tissue with a time-to-maximum (Tmax) >10 seconds to that with Tmax >6 seconds in perfusion MRI (Magnetic resonance image). In this study, using perfusion MRI, we hypothesized that HIR could be associated with intracranial atherosclerotic disease (ICAD) in acute MCA occlusion. We also performed machine learning-based feature importance analysis to identify factors associated with the mechanism of acute MCA occlusion. Methods We analyzed 117 patients with acute MCA occlusion treated with EVT and underwent RAPID MRI between March 2020 and December 2023. Patients were classified into the ICAD and non-ICAD groups. Clinical and imaging parameters were assessed using logistic regression. The cutoff value of HIR was calculated using Youden's index, and its predictive value was compared using the DeLong test. Development of a machine learning algorithm for predicting ICAD using XGBoost (eXtreme Gradient Boosting) and feature importance analysis were performed. Results 34% were ICAD group with higher low-density lipoprotein cholesterol (LDL-C) levels, mild NIHSS, small DWI size, and low HIR. Late OTP (> 6hours) (OR 9.67), no SVS (OR 9.05), no initial AF (OR 63.67), low HIR (<0.31) (OR 7.70) were associated with ICAD. Including HIR improved predictive performance (AUC 0.73 vs 0.88, p-value < 0.001). The XGBoost model showed high accuracy (0.92), SHAP value identified HIR, OTP, initial AF, SVS were important factors. Conclusion HIR is a potential marker to determine the mechanism of acute MCA occlusion. Combined with established clinical and radiological features, it helps in identifying the ICAD-related MCA occlusion before EVT.