A novel method of artery stenosis diagnosis using transfer function and support vector machine based on transmission...

Area of Science:

• Biomedical Engineering
• Cardiovascular Physiology
• Computational Medicine

Background:

• Arterial stenosis significantly impacts hemodynamics.
• Transfer function (TF) is crucial for analyzing blood flow in diseased arteries.
• Accurate diagnosis of arterial stenosis is vital for patient outcomes.

Purpose of the Study:

• To validate the feasibility of using TF for diagnosing arterial stenosis.
• To simulate forward and inverse problems related to arterial stenosis using TF.
• To develop a diagnostic model for arterial stenosis.

Main Methods:

• A 55-segment transmission line model (TLM) of the human arterial tree was used to calculate TF.
• The influence of stenosis degree (10-90%) and location in major arteries (carotid, aorta, iliac) on TF was analyzed.
• A TF database was created using the TLM model to train a support vector machine (SVM) for stenosis diagnosis.

Main Results:

• TF modulus and phase showed sharp decreases with increasing stenosis degree from 2-10Hz.
• Nonlinear TF characteristics were observed at frequencies above 10Hz.
• The SVM model achieved an average diagnostic accuracy above 76% for stenosis degrees from 10% to 90%.

Conclusions:

• The combined TF and SVM approach is a theoretically feasible method for diagnosing arterial stenosis.
• High diagnostic accuracies were achieved for moderate (87%) and severe (99%) stenosis.
• Stenosis localization accuracy reached up to 94% for 90% stenosis.