The value of radiomics features derived from cardiac MR cine images in predicting late gadolinium enhancement in patients with hypertrophic cardiomyopathy
10.3760/cma.j.cn112149-20230109-00026
- VernacularTitle:心脏MR平扫电影图像的影像组学特征对肥厚型心肌病患者延迟强化的预测价值
- Author:
Hongbo ZHANG
1
;
Lei ZHAO
;
Haoru WANG
;
Guanyu LU
;
Chen ZHANG
;
Guoxi XIE
;
Na LU
;
Xiaohai MA
Author Information
1. 首都医科大学附属北京安贞医院介入诊疗科，北京 100029
- Keywords:
Cardiomyopathy, hypertrophic;
Magnetic resonance imaging;
Radiomics;
Late gadolinium enhancement
- From:
Chinese Journal of Radiology
2023;57(11):1231-1238
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To investigate the value of radiomics features derived from cardiac MR (CMR) cine images for predicting late gadolinium enhancement (LGE) in patients with hypertrophic cardiomyopathy (HCM).Methods:Firstly, a total of 300 HCM patients with definite diagnosis who underwent CMR examination in Beijing Anzhen Hospital from May 2017 to August 2021 were retrospectively included, and were divided into a training set and a test set with a proportion of 7∶3 using random stratified sampling method. Then, a total of 89 HCM patients with definite diagnosis who underwent CMR examination in Beijing Anzhen Hospital from January 2022 to May 2023 were included for external validation. The CVI42 software was used to obtain the cardiac function parameters. Intraclass correlation coefficient (ICC), Pearson correlation coefficient and least absolute shrinkage and selection operator (LASSO) were used to select radiomics features. Finally, LASSO regression and three machine learning algorithms (including support vector machine, linear discriminant analysis and naive Bayes) were used to build prediction models. The area under the receiver operating characteristic curve (AUC) was used to evaluate the prediction value of the model.Results:Totally 1 409 features were extracted from each patient, and 19 features were retained to build radiomics signature after dimension reduction. Although no significant differences among the four methods, the prediction performance and stability of LASSO regression were relatively good. The AUC was 0.795 (95%CI 0.735-0.855) in the training set, 0.765 (95%CI 0.668-0.862) in the test set and 0.721(95%CI 0.598-0.845) in the external validation set.Conclusions:The features extracted from CMR cine images can be used to predict LGE in HCM patients. LASSO regression is recommended for model construction.
