Automated plaque classification using computed tomography angiography and Gabor transformations.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Cardiology

Background:

• Cardiovascular diseases, particularly coronary artery disease (CAD), are a leading global cause of mortality.
• Atherosclerosis-induced inflammation in coronary arteries (CA) can lead to CAD, with coronary artery calcification (CAC) being a key predictor.
• Computed tomography angiography (CTA) is a vital non-intrusive imaging tool for characterizing CA plaques.

Purpose of the Study:

• To develop and evaluate an automated algorithm for classifying coronary artery plaques as normal, calcified, or non-calcified using CTA images.
• To assess the efficacy of Gabor transform-based features and various classification methods for plaque characterization.

Main Methods:

• An automated algorithm was developed using 2646 CTA images from 73 patients.
• Seven features (energy, Kapur, Max, Rényi, Shannon, Vajda, and Yager entropies) were extracted from Gabor transform coefficients.
• Features were ranked using F-value, and classification was performed using various methods, including a probabilistic neural network, with and without feature reduction techniques.

Main Results:

• The automated algorithm achieved high classification performance without feature reduction.
• A probabilistic neural network utilizing all computed Gabor features yielded the best results.
• Performance metrics included 89.09% accuracy, 91.70% positive predictive value, 91.83% sensitivity, and 83.70% specificity.

Conclusions:

• The developed automated technique demonstrates significant potential for classifying coronary artery plaques in CTA images.
• This method could serve as a valuable tool for clinicians in plaque diagnostics, potentially reducing procedural costs and patient radiation dose.