Classification of carotid artery Doppler signals in the early phase of atherosclerosis using complex-valued...

Area of Science:

• Biomedical Engineering
• Medical Imaging Analysis
• Artificial Intelligence in Medicine

Background:

• Atherosclerosis is a significant cardiovascular disease.
• Early detection of atherosclerosis is crucial for effective management.
• Carotid Doppler ultrasound provides valuable hemodynamic information.

Purpose of the Study:

• To evaluate the effectiveness of complex-valued artificial neural networks (CVANN) for classifying early-stage atherosclerosis.
• To enhance CVANN performance using Principal Component Analysis (PCA) for feature extraction and Fuzzy C-Means (FCM) clustering for data reduction.
• To assess the diagnostic accuracy of PCA-CVANN and FCM-CVANN architectures in differentiating healthy individuals from patients with early atherosclerosis.

Main Methods:

• Acquisition of carotid arterial Doppler ultrasound signals from 38 patients with early atherosclerosis and 40 healthy volunteers.
• Development of two CVANN architectures: PCA-CVANN (incorporating PCA for feature extraction) and FCM-CVANN (incorporating FCM for dataset reduction).
• Random selection of training and testing data using 10-fold cross-validation.

Main Results:

• Both PCA-CVANN and FCM-CVANN architectures achieved approximately 100% correct classification rates for both training and testing datasets.
• The developed models successfully distinguished between healthy subjects and patients diagnosed with early-stage atherosclerosis.
• Feature extraction via PCA and data reduction via FCM significantly improved the effectiveness of the CVANN system.

Conclusions:

• PCA-CVANN and FCM-CVANN are highly effective methods for classifying carotid Doppler ultrasound signals.
• These AI-driven approaches demonstrate excellent potential for the accurate and early diagnosis of atherosclerosis.
• The study highlights the successful application of advanced machine learning techniques in cardiovascular diagnostics.