Related Experiment Video
Updated: Jun 27, 2025
Quantification of Mouse Heart Left Ventricular Function, Myocardial Strain, and Hemodynamic Forces by Cardiovascular Magnetic Resonance Imaging
Published on: May 24, 2021
Integration of Cine-cardiac Magnetic Resonance Radiomics and Machine Learning for Differentiating Ischemic and
Jia Deng1, Langtao Zhou2, Yueyan Li3
1The First Affiliated Hospital, Department of Radiology, Hengyang Medical School, University of South China, Hengyang, Hunan 421001, China (J.D., B.L., G.L., H.Z.); The First Affiliated Hospital, Department of Cardiology, Hengyang Medical School, University of South China, Hengyang, Hunan 421001, China (J.D., Y.L., Y.Y., J.Z., H.T.).
Machine learning models using cardiac MRI radiomics can differentiate ischemic cardiomyopathy (ICM) from dilated cardiomyopathy (DCM). Radiomics models, particularly those using myocardium images, show superior diagnostic potential compared to clinical features alone.
Area of Science:
- Cardiology
- Medical Imaging
- Machine Learning
Background:
- Distinguishing between ischemic cardiomyopathy (ICM) and dilated cardiomyopathy (DCM) is crucial for effective patient management.
- Current diagnostic methods may have limitations in accuracy and invasiveness.
Purpose of the Study:
- To evaluate the capability of machine learning algorithms in utilizing radiomic features from cine-cardiac magnetic resonance (CMR) sequences for differentiating ICM from DCM.
- To compare the diagnostic performance of radiomics-based models against clinical feature-based models.
Main Methods:
- A retrospective study of 115 cardiomyopathy patients (64 ICM, 51 DCM).
- Extraction of radiomic features from left ventricle, LV cavity, and myocardium regions of interest in cine-CMR sequences.
- Testing of 10 classical machine learning classifiers with fivefold cross-validation.
- Comparison of models based on invasive clinical (IC), noninvasive clinical (NIC), and combined clinical (CC) features versus radiomics features.
Main Results:
- In validation, Gaussian naive Bayes (GNB) models achieved high AUCs for clinical features (0.879-0.906).
- The myocardium (MYO) radiomics model (MYO_LASSOCV_MLP) showed the highest AUC (0.919) in validation.
- In the test set, a specific radiomics model (MYO_RFECV_GNB) achieved an AUC of 0.857, outperforming clinical models (AUCs 0.732-0.786).
Conclusions:
- Radiomics models utilizing myocardium images from cine-CMR show significant potential for differentiating ICM from DCM.
- Integration of radiomics and machine learning enhances diagnostic capability, potentially reducing examination risks and duration.
