High-Dimensional Feature Selection for Automatic Classification of Coronary Stenosis Using an Evolutionary Algorithm.

Area of Science:

• Medical Imaging
• Computational Biology
• Machine Learning

Background:

• Coronary stenosis classification is crucial for cardiovascular disease management.
• High-dimensional feature selection poses a significant challenge in medical image analysis.
• Existing methods struggle with the complexity of automated coronary stenosis detection.

Purpose of the Study:

• To develop a novel evolutionary algorithm for high-dimensional feature selection in coronary stenosis classification.
• To identify a minimal yet effective subset of features for accurate disease detection.
• To evaluate the proposed method against state-of-the-art techniques.

Main Methods:

• A feature extraction stage generated 473 features (intensity, texture, shape).
• An evolutionary algorithm performed feature selection on the high-dimensional feature bank (O(2^473) search space).
• A Support Vector Machine (SVM) classifier was trained and validated using the selected feature subset.

Main Results:

• A four-feature subset achieved a 99% discrimination rate.
• The four-feature subset yielded high classification performance: 0.86 accuracy and 0.75 Jaccard coefficient on the primary dataset.
• On a larger public dataset (2788 instances), the method achieved 0.89 accuracy and 0.80 Jaccard Coefficient.

Conclusions:

• The proposed evolutionary feature selection strategy effectively identifies a small, discriminative feature subset for coronary stenosis.
• The identified four-feature subset demonstrates high classification performance, suitable for clinical decision support systems.
• This approach offers a promising avenue for automated and accurate diagnosis of coronary artery disease.