Comparing Classifiers for the Prediction of the Stenosis of Coronary Artery

Abstract

Myocardial Ischemia is the main cause of mortality in patients with Coronary Artery Disease (CAD). One of the methods used in screening patients with this disease is the diag-nosis of radionuclide myocardial perfusion imaging (rMPI). In this paper, we conducted a comparative study by experimenting on several machine learning models, such as Logistic Regression, Random Forest, XGBoost, etc., to classify the stenosis of coronary artery. High-level features from rMPI computed by 4D-MSPECT polar map were used to train/test the models. rMPI features of the risk group of CAD patients were obtained from a public hospital. With the hypothesis that patient characteristics (e.g., Diabetes Mellitus, Hypertension, Dyslipidemia) could improve the prediction performance of the models, this study also included patient characteristics in our experimentation as important parts of feature selection. All other processes (i.e., data cleaning, feature selection, feature engineering and feature transformation) in machine learning pipeline were also deliberately experimented in this study. For model selection, two-level validation regarding generalization and hyperparameter tuning were also performed.