Enhancing Prediction of Myocardial Recovery After Coronary Revascularization: Integrating Radiomics from Myocardial
Deyi Huang1, Xingan Yang2, Hongbiao Ruan3
1Department of Ultrasound, The People's Hospital of Yuhuan, Yuhuan City, Zhejiang Province, People's Republic of China.
International Journal of General Medicine
|June 6, 2024
Summary
Machine learning and radiomics integrated with myocardial contrast echocardiography accurately predict functional recovery in coronary artery disease patients. This approach enhances chronic CAD management by precisely identifying myocardial recovery potential.
Area of Science:
- Cardiology
- Medical Imaging
- Artificial Intelligence
Background:
- Chronic coronary artery disease (CAD) management relies on myocardial contrast echocardiography (MCE), but interpretation is subjective and distinguishing hibernating from necrotic myocardium is challenging.
- Predicting functional recovery in dysfunctional myocardial segments after revascularization is crucial for optimizing patient outcomes.
Purpose of the Study:
- To explore the integration of machine learning (ML) with radiomics from MCE to predict functional recovery in dyskinetic myocardial segments in CAD patients.
- To overcome the limitations of subjective MCE interpretation and improve the accuracy of predicting myocardial recovery.
Main Methods:
- A prospective study of 55 chronic CAD patients, divided into training and testing sets.
- Radiomics features were extracted from MCE images of dysfunctional myocardial segments.
- Four ML classifiers were trained and compared, integrating radiomics features and MCE parameters to predict functional recovery.
Main Results:
- Myocardial blood flow (MBF) was the most precise clinical predictor of recovery (AUC 0.770).
- Nine radiomics features were identified as key predictors.
- A random forest (RF) model integrating MBF and radiomics features achieved superior predictive accuracy (AUC 0.821 on testing data).
Conclusions:
- ML integrated with MCE-derived radiomics effectively predicts myocardial recovery in CAD.
- The RF model offers a non-invasive, precise approach for enhanced CAD management.
