Classification of Carotid Plaque Echogenicity by Combining Texture Features and Morphologic Characteristics

Area of Science:

• Vascular Ultrasound
• Medical Imaging Analysis
• Cardiovascular Disease Prediction

Background:

• Anechoic carotid plaques on sonography are indicators of future cardiovascular events.
• Objective assessment of plaque echogenicity is needed for early prediction.
• Texture and morphologic features may offer a promising method.

Purpose of the Study:

• To objectively assess carotid plaque echogenicity.
• To combine texture features from MaZda software and morphologic characteristics.
• To evaluate the method's potential for early cardiovascular disease prediction.

Main Methods:

• Collected 268 plaque images from 136 volunteers, classifying them into hyperechoic, intermediate, and anechoic.
• Extracted ~300 texture features using MaZda software algorithms.
• Measured morphologic characteristics (stenosis, intima-media thickness, plaque length) via B-mode sonography.

Main Results:

• Identified 30 statistically significant features and 2 most discriminative features.
• Achieved classification accuracy rates of 72.03% and 88.14% for plaque types.
• Demonstrated an area under the curve of 0.918 for identifying anechoic plaques.

Conclusions:

• Classifying carotid plaque echogenicity is feasible using combined texture and morphologic features.
• This objective approach may enhance early cardiovascular disease risk stratification.
• Further validation could establish this as a valuable diagnostic tool.