MO782: Evaluation of Acceleromet Reuse for Method of Measurement for Turbulent Blood Flow in Arteriovenous Fistulas in Haemodialysis Patients

Abstract

Abstract BACKGROUND AND AIMS Haemodialysis (HD) is the primary treatment for end-stage renal disease. HD therapy requires access to the vascular system. This can be achieved through surgically constructing an arteriovenous fistula (AVF). Post-surgery up to 30% of patients experience dysfunction of the AVF, which may lead to a chronically damaged AVF and render the patient without HD treatment [1]. At present no satisfactory post-surgery monitoring of the AVF exist and, therefore, malfunction of the AVF is recognized too late at a time where no rescue is possible. The functionality of the AVF is normally assessed by auscultation of the murmurs produced by the blood flow. Narrowing of the AVF caused by thrombosis or stenosis causes changes in the blood flow, thus causing changes in the murmurs [2]. The changes in murmurs have not been widely studied but may be of importance to gain knowledge on AVF malfunction. This could be used in developing a method for monitoring the AVF post-surgery. METHOD Acoustic recordings was obtained from 21 patient post AVF-surgery at Aalborg University Hospital. Recordings were made with a Silicon Designs 1521 accelerometer encapsulated in a 3D printed box, attached to the skin using double adhesive tape 2–5 cm proximal to the AVF. A total of 11 patients were recorded twice resulting in a total of 32 recordings. Data were sampled at 5000Hz. All data processing and analysis were performed using MATLAB 2021b (MathWorks Inc.). Every recording was manually analysed and labelled with or without audible murmurs. Before processing every recording was split into segments of 3 sec length resulting in a total of 697 segments. Data was processed with a signal processing method [wavelet packet decomposition (Daubechies 4)] to analyse if the frequencies of the sounds differed between recordings with and without murmurs. Energy of the wavelet scales (Scale Energy) was calculated for each frequency band. The scale energy was then log-transformed using natural logarithm to make data normally distributed, and converted to decibel (dB). RESULTS A total of 654 segments contained murmurs and 43 segments did not contain murmurs. In all frequency bands, murmurs resulted in a significant change in energy level (P < 0.05). CONCLUSION This study shows AVF murmurs results in a significant change in acoustic energy across all frequency bands. This demonstrates that accelerometers can be used to quantify murmurs from AVFs post-surgery.