Myocardial Function Prediction After Coronary Artery Bypass Grafting Using MRI Radiomic Features and Machine Learning...

Area of Science:

• Cardiology
• Radiology
• Data Science

Background:

• Coronary artery bypass grafting (CABG) is a common procedure for heart disease.
• Predicting myocardial function improvement post-CABG is crucial for patient management.
• Cardiac Magnetic Resonance (CMR) imaging, particularly late gadolinium enhancement (LGE-CMR), provides valuable insights into myocardial scar and function.

Purpose of the Study:

• To predict myocardial function improvement in patients after CABG using radiomics and machine learning on LGE-CMR images.
• To evaluate the efficacy of radiomics features in classifying CABG responders versus non-responders.
• To identify key radiomics features that correlate with functional recovery after CABG.

Main Methods:

• 43 patients undergoing CABG with visible LGE-CMR scars were included.
• Preoperative LGE-CMR images were analyzed using radiomics feature extraction after resampling and intensity quantization.
• Machine learning algorithms, including SCAD-penalized SVM and Recursive Partitioning (RP), were employed for classification and validated using cross-validation and bootstrapping.

Main Results:

• Machine learning models, particularly SCAD-penalized SVM, demonstrated significant predictive performance.
• SCAD-penalized SVM achieved an AUC of 0.784 for CABG responder classification.
• Specific radiomics features, like GLSZM gray-level non-uniformity-normalized, showed prognostic value, especially in multivariable analysis.

Conclusions:

• Radiomics features, analyzed with machine learning algorithms, provide valuable prognostic information for myocardial function after CABG.
• Multivariate analysis combining radiomics features enhances the prediction of functional recovery post-CABG.
• This approach aids in assessing patient outcomes and optimizing treatment strategies.