Prediction of Early Perihematomal Edema Expansion Based on Noncontrast Computed Tomography Radiomics and Machine Learning in Intracerebral Hemorrhage

作者全名："Li, Yu-Lun; Chen, Chu; Zhang, Li-Juan; Zheng, Yi-Neng; Lv, Xin-Ni; Zhao, Li-Bo; Li, Qi; Lv, Fa-Jin"

作者地址："[Li, Yu-Lun; Zhang, Li-Juan; Zheng, Yi-Neng; Lv, Fa-Jin] Chongqing Med Univ, Dept Radiol, Affiliated Hosp 1, Chongqing, Peoples R China; [Chen, Chu; Lv, Xin-Ni; Li, Qi] Chongqing Med Univ, Dept Neurol, Affiliated Hosp 1, Chongqing, Peoples R China; [Li, Qi] Anhui Med Univ, Dept Neurol, Affiliated Hosp 2, Hefei, Peoples R China; [Zhao, Li-Bo; Li, Qi] Chongqing Key Lab Cerebrovasc Dis Res, Chongqing, Peoples R China; [Zhao, Li-Bo] Chongqing Med Univ, Dept Neurol, Yongchuan Hosp, Chongqing, Peoples R China"

通信作者："Lv, FJ (通讯作者)，Chongqing Med Univ, Dept Radiol, Affiliated Hosp 1, Chongqing, Peoples R China.; Li, Q (通讯作者)，Chongqing Med Univ, Dept Neurol, Affiliated Hosp 1, Chongqing, Peoples R China.; Li, Q (通讯作者)，Anhui Med Univ, Dept Neurol, Affiliated Hosp 2, Hefei, Peoples R China.; Li, Q (通讯作者)，Chongqing Key Lab Cerebrovasc Dis Res, Chongqing, Peoples R China."

来源：WORLD NEUROSURGERY

ESI学科分类：CLINICAL MEDICINE

WOS号：WOS:001029786800001

JCR分区：Q2

影响因子：1.9

年份：2023

卷号：175

期号：　

开始页：E264

结束页：E270

文献类型：Article

关键词：Brain edema; Cerebral hemorrhage; Machine learning; Radiomics; Tomography

摘要："- OBJECTIVES: To investigate the predictive value of nonradiomics features and machine learning for early perifrom 214 patients with spontaneous ICH. All radiomics features were extracted from volume of interest of hematomas on admission scans. A total of 8 machine learning methods were applied for constructing models in the training and the test set. Receiver operating characteristic analysis and the areas under the curve were used to evaluate the predictive value.- RESULTS: A total of 23 features were finally selected to establish models of early PHE expansion after feature screening. Patients were randomly assigned into training (n = 171) and test (n = 43) sets. The accuracy, sensitivity, and specificity in the test set were 72.1%, 90.0%, and 66.7% for the support vector machine model; 79.1%, 70.0%, and 84.4% for the k-nearest neighbor model; 88.4%, 90.0%, and 87.9% for the logistic regression model; 74.4%, 90.0%, and 69.7% for the extra tree model; 74.4%, 90.0%, and 69.7% for the extreme gradient boosting model; 83.7%, 100%, and 78.8% for the multilayer perceptron (MLP) model; 72.1%, 100%, and 65.6% for the light gradient boosting machine model; and 60.5%, 90.0%, and 53.1% for the random forest model, respectively.- CONCLUSIONS: The MLP model seemed to be the best model for prediction of PHE expansion in patients with ICH. NCCT models based on radiomics features and machine learning could predict early PHE expansion and improve the discrimination of identify spontaneous intracerebral hemorrhage patients at risk of early PHE expansion."

基金机构："National Natural Science Foundation of China [82071337]; National Key Research and Development Program of China [2018YFC1312200, 2018YFC1312203]; Chongqing High-end Young Investigator Project [2019GDRC005]"

基金资助正文："This study was supported bygrants from the National Natural Science Foundation of China (no. 82071337), the National Key Research and Development Program of China (grant no. 2018YFC1312200,no. 2018YFC1312203), and the Chongqing High-end Young Investigator Project (no. 2019GDRC005)."