A Novel Classification Technique of Arteriovenous Fistula Stenosis Evaluation Using Bilateral PPG Analysis

Area of Science:

• Biomedical Engineering
• Medical Diagnostics
• Vascular Access Monitoring

Background:

• Hemodialysis (HD) is a common treatment for end-stage renal disease (ESRD).
• Arteriovenous fistula (AVF) is crucial for HD access, but stenosis frequently causes dysfunction.
• Accurate and noninvasive stenosis evaluation is needed for AVF monitoring.

Purpose of the Study:

• To propose a noninvasive method for evaluating arteriovenous fistula stenosis (AVF) using photoplethysmography (PPG).
• To assess the performance of an error correcting output coding support vector machine one versus rest (ESVM-OVR) classifier for degree of stenosis (DOS) evaluation.
• To compare the proposed PPG-based system with an artificial neural network (ANN) classifier.

Main Methods:

• Asynchronous analysis of bilateral photoplethysmography (PPG) signals.
• Implementation of an error correcting output coding support vector machine one versus rest (ESVM-OVR) classifier for degree of stenosis (DOS) evaluation.
• Data collection from 22 patients' thumbs, with comparison against an artificial neural network (ANN) classifier.

Main Results:

• The proposed PPG-based system achieved a positive predictive value (PPV) of 91.67% for AVF stenosis evaluation.
• The system demonstrated superior noise tolerance compared to other methods.
• The ESVM-OVR classifier outperformed the ANN classifier in this diagnostic task.

Conclusions:

• The noninvasive PPG analysis with ESVM-OVR offers a promising approach for evaluating AVF stenosis.
• The system shows potential for integration into wearable devices for continuous AVF monitoring.
• This method could aid in the early detection and management of AVF dysfunction in hemodialysis patients.