Investigations on coronary artery plaque detection and subclassification using machine learning classifier.

Area of Science:

• Medical Imaging
• Machine Learning
• Cardiovascular Disease Research

Background:

• Coronary artery disease (CAD) poses a significant health risk due to arterial plaque buildup.
• Accurate plaque detection and classification are crucial for reducing patient mortality.

Purpose of the Study:

• To develop a novel clustering method for plaque segmentation.
• To extract features using wavelet transform for plaque classification.
• To evaluate machine learning classifiers for CAD plaque analysis.

Main Methods:

• Plaque segmentation using a novel clustering approach.
• Feature extraction via wavelet transform.
• Classification using Support Vector Machine, Random Forest, and Decision Tree models.
• Ensemble classification with a bootstrap voting method.

Main Results:

• Achieved a classification accuracy of 97.7%.
• Reported Sensitivity and Specificity rates of 97.8% and 97.5%, respectively.
• Demonstrated high performance on a dataset of 64 normal and 73 abnormal CTA images for training, and 111 normal and 103 abnormal CTA images for testing.

Conclusions:

• The proposed image processing and machine learning scheme is feasible and advantageous for CAD plaque detection and classification.
• This method can assist in improving patient outcomes for coronary artery diseases.
• Highlights the potential of AI in cardiovascular diagnostics.