CT纹理分析联合机器学习对急性脑梗死出血性转化的预测价值分析

Predictive Value of CT Texture Analysis Combined with Machine Learning in Hemorrhagic Transformation of Acute Cerebral Infarction

目的探究CT纹理分析联合机器学习对急性脑梗死后出血性转化的预测价值.方法回顾性分析2021年1月至2023年9月入院治疗的急性脑梗死患者的CT图像资料,比较溶栓治疗后发生出血性转化组(n=78)和未出血组(n=122)之间CT梗死区形态参数的差异,在CT图像上提取梗死区域的纹理特征参数,分别基于纹理、形态特征构建多种机器学习模型,采用受试者工作特征(ROC)曲线及曲线下面积(AUC)评估模型的预测效能。结果梗死部位梗死面积及是否为多发梗死灶等形态特征具有统计学意义(P<0.05);以纹理特征构建的机器学习模型能更好的预测出血性转化,整体效能高于形态特征模型,其中XGBoost和CatBoost预测效能最高。结论基于CT纹理分析可有效预测急性脑梗死出血性转化。

Objective To explore the predictive value of CT texture analysis combined with machine learning in hemorrhagic transformation after acute cerebral infarction.Methods The CT image data of patients with acute cerebral infarction admitted to hospital from January,2021 to September,2023 were retrospectively analyzed,and the differences of morphological parameters of CT infarcted area between hemorrhagic transformation group(n=78)and non-hemorrhagic group(n=122)after thrombolytic therapy were compared.The texture feature parameters of infarcted area were extracted from CT images,and various machine learning models were constructed based on the texture and morphological features,respectively.Results The morphological characteristics of the infarct area and whether it was multiple infarcts were statistically significant(P<0.05).The machine learning model based on texture features can better predict hemorrhagic transformation,and its overall efficiency is higher than that of morphological feature models,among which XGBoost and CatBoost have the highest prediction efficiency.Conclusion Texture analysis based on CT can effectively predict hemorrhagic transformation of acute cerebral infarction.