Leveraging Acoustic Analysis to Detect Arteriovenous Fistula Complications
Open Access
- 10 June 2025
- journal article
- Published by Wolters Kluwer Health in Kidney360
- Vol. 6 (10) , 1771-1779
Abstract
Background: Vascular access failure often makes hemodialysis treatment inefficient or impossible. This study investigates the potential of using recordings and acoustic analysis of arteriovenous fistula (AVF) sounds to i) distinguish between AVFs suitable for routine cannulation and those that are unsuitable and ii) detect possible complications in functional AVFs early. Methods: A total of 154 AVF sounds were recorded and classified into functioning and non-functioning groups. Functioning AVFs were further categorized as optimally patent or as having complications like stenosis or aneurysm, based on ultrasound. After audio pre-processing, the high-low peak ratio (HLPR) and its natural logarithm (ln(HLPR)) were calculated using peak amplitudes in low and high-frequency ranges (100-250 Hz and 500-700 Hz respectively). Results: The results revealed significant differences in acoustic metrics between functioning and non-functioning AVFs. Non-functioning AVFs exhibited higher frequency sounds, higher HLPR and ln(HLPR), and reduced amplitude in the low frequency range. In functioning AVFs, the presence of stenosis produced distinct acoustic patterns with higher frequencies and lower amplitude in the low-frequency range, while aneurysms exhibited prominent high frequencies but did not significantly affect acoustic metrics compared to sound recordings from optimally patent AVFs. Optimally patent AVFs had the highest amplitude in low frequency and the lowest high-frequency presence, exhibiting statistically significant differences compared to stenotic and non-functioning AVFs. Additionally, exploratory ROC analysis suggests a potential threshold in ln(HLPR) for identifying AVF function impairment, with good sensitivity and specificity. Conclusions: This study highlights the transformative potential of AVF sound analysis as a tool for identifying AVF complications and failure. By using the unique acoustic characteristics of AVFs, clinicians can improve their surveillance strategies, providing patients with timely interventions and better outcomes.Keywords
This publication has 15 references indexed in Scilit:
- The prototype device for non-invasive diagnosis of arteriovenous fistula condition using machine learning methodsScientific Reports, 2020
- Evaluation of Hemodialysis Arteriovenous Bruit by Deep LearningSensors, 2020
- KDOQI Clinical Practice Guideline for Vascular Access: 2019 UpdateAmerican Journal of Kidney Diseases, 2020
- How to meet the demand for good quality renal dialysis as part of universal health coverage in resource-limited settings?Health Research Policy and Systems, 2016
- Vascular Access Choice in Incident Hemodialysis PatientsJournal of the American Society of Nephrology, 2015
- Novel Noninvasive Approach for Detecting Arteriovenous Fistula StenosisIEEE Transactions on Biomedical Engineering, 2014
- Computer assisted Sound Analysis of Arteriovenous Fistula in Hemodialysis PatientsThe International Journal of Artificial Organs, 2013
- Interventional NephrologyClinical Journal of the American Society of Nephrology, 2013
- Clinical evaluation of the 3M Littmann Electronic Stethoscope Model 3200 in 150 catsJournal of Feline Medicine and Surgery, 2013
- Computerised analysis of auscultatory sounds associated with vascular patency of haemodialysis accessMedical & Biological Engineering & Computing, 2005