Prognosis Evaluation Of Acute Ischemic Stroke In Anterior Circulation

Part 1 A preliminary study of brain-stem white matter hyperintensity on FLAIR in evaluating the prognosis of anterior circulation AISObjective: Previous studies have shown that supratentorial white matter hyperintensity on FLAIR is related to the outcome of acute ischemic stroke,but the effect of brain-stem white matter hyperintensity(BWMH)on FLAIR is not clear.The purpose of this study was to explore the value of BWMH in evaluating the outcome of anterior circulation AIS.Methods: The clinical and routine MR data of anterior circulation AIS from 2019.1 to 2020.9 were collected retrospectively.The diagnosis of AIS was based on the high signal lesions on DWI in the anterior circulation with the clinical manifestations of acute stroke.The mRS>2 in 90 days was regarded as poor outcome.Logistic regression analysis was used to compare the clinical and routine MR factors of patients with poor outcome.Draw the receiver operating characteristic curve(ROC)to analyze different models and the comprehensive model including brain-stem white matter hyperintensity on FLAIR to evaluate the efficacy of clinical prognosis.Results: Among 159 patients with AIS,there were 114 males(71.70%) and 45 females(28.30%),with an average age of(60.07±11.88)years.BWMH was positive in 57 cases(35.85%)and negative in102cases(65.15%).Among the 112 cases with good outcome,28 cases were BWMH positive(25%),and 29 cases were BWMH positive(61.70%)in 47 cases with poor outcome.Logistic regression analysis showed that BWMH(OR=4.191,95%CI: 1.606,10.936;P= 0.003)was an independent prognostic factor with poor outcome in 90 days.There were significant differences in age(P<0.001),admission NIHSS score(P=0.004)and Fazekas score(P<0.001)between the two groups.ROC showed that the area under the curve(AUC)of conventional MR model and clinical model were 0.759 and 0.839 respectively,and AUC increased to 0.796 and 0.863 when BWMH factor was added to the above two models,while the AUC of comprehensive model(routine MR+clinical factor + BWMH)was 0.877 with sensitivity,specificity and Yoden index 0.851,0.83 and 0.681,respectively.Conclusions: BWMH is a valuable biological biomarke for poor outcome of anterior circulation AIS.This sign is helpful to improve the efficacy of clinical imaging in predicting the prognosis of patients with anterior circulation AIS.Part 2 FALIR and DWI-based radiomics features as prognostic biomarkers for anterior circulation acute ischemic stroke patients:a preliminary studyObjective: At present,it is still challenging to predict the clinical outcome of acute ischemic stroke(AIS)only based on the information from clinical and conventional magnetic resonance imaging(MRI).This study aimed at developing radiomics features based on fluid-attenuated inversion recovery(FLAIR)and diffusion-weighted imaging(DWI)as a prognostic biomarkers,and to validate the improvement of these radiomics features when adding to combined model to predict the clinical outcomes of anterior circulation AIS patients.Methods: In total,190 patients with anterior circulation AIS,including training cohort(n=110)and external testing cohort(n=80)were retrospectively included.We selected the clinical and conventional MRI factors which influence the outcome of AIS with Logistic regression analysis.The Insight Segmentation and Registration Toolkit-ANAP(ITK-SNAP)was employed to segment AIS lesions in FLAIR and DWI images.We selected radiomics features with a flexible open-source platform,the Pyradiomics.The strongest radiomics features were screened out after dimension reduction by Lasso regression.Then we setup several outcome predicting models and evaluated their diagnostic efficacy with receiver operating characteristic curve(ROC).Results: The clinical and conventional factors which related to unfavorable outcome included age(>60.5-year-old),gender(female),orthogonal diameter(OD),admission NIHSS score,and admission DWI-ASPECTS score(P<0.05).Multivariate Logistic analysis revealed that age(OR=1.070,95%CI 1.034-1.107),gender(OR=2.974,95%CI 1.299-6.810),admission NIHSS score(OR=1.270,95%CI1.156-1.394)were the independent predictor of unfavorable outcome for AIS patients.The four strongest radiomics features which related to unfavorable outcome were DWI original GLCM Joint Energy,DWI Wavelet HL GLSZM Gray Level Nonuniformity Normalzied,DWI log sigma 3.0 mm 3D GLSZM Gray Level Variance,and FLAIR original GLSZM Zone Entropy.For predicting unfavorable outcome of AIS in training cohort,the area under ROC(AUC)of DWI radiomics model was 0.766,FLAIR radiomics model 0.736,DWI and FLAIR radiomics model 0.802,clinical and conventional MRI model 0.880;for external testing cohort,AUC for the prediction with DWI radiomics model was 0.772,FLAIR radiomics model 0.759,DWI and FLAIR radiomics model 0.803,clinical and conventional MRI model 0.755.However,when adding radiomics features to the combined model,the AUCs for predicting unfavorable outcome in training and external testing cohorts reached 0.911 and 0.810 separately.Conclusions: The radiomics features based on FLAIR and DWI are the prognostic biomarkers for the unfavorable clinical outcome of anterior circulation AIS.The predictive performance would be improved when adding radiomics features to combined prediction model.