Ischemic lesion volume prediction in thrombolysis treated wake-up stroke patients

Abstract

There is growing research interest on identification of CT Perfusion (CTP) parameters that predict the outcome in acute ischemic stroke patients. The aim of this study is to produce the model, based on core-penumbra related parameters assessed by CTP processing and commonly used clinical prediction factors, to predict the final infarct volume in thrombolysis-treated anterior circulation wake-up stroke (WUS) patients. The study was conducted on 51 consecutive wake-up ischemic stroke patients. The model for the predictive estimation of final ischemic volume was determined by using the Least Absolute Shrinkage and Selection Operator (LASSO) regularized least-squares regression. The results showed that CTP core volume and CTP total ischemic volume at admission, together with ASPECT score predict the final infarct lesion volume. In particular, the identified model presented 5-fold cross-validation root mean square error RMSE of 8.1 ml and the coefficient of determination (R2=0.94) on our dataset. The results should be confirmed in a lager study. In conclusion, in this study we preliminarily identified a predictive model to estimate final ischemic lesion volume in thrombolysis treated wake-up stroke patients.