VANet: A Solution for Ventricular Arrhythmias Detection of IEGM on Embedded Devices

Abstract

Real time detection of ventricular arrhythmias (VAs) in patients and timely provision of defibrillation treatment are crucial in the recording of intracardiac electrograms (IEGMs). Recently, deep convolutional networks have been used to detect VAs in IEGM recordings. However, due to their complex computations and model structures make them difficult to deploy on resource-constrained, low-power embedded devices. This letter introduces VANet, a fully convolutional lightweight neural network architecture for detecting VAs in IEGM recordings. Our hardware INT8 quantization implementation method effectively enables its deployment on embedded devices. Results show that it achieves the best performance in terms of accuracy, storage, and latency compared to state-of-the-art network architectures, X-Cube-AI and Tensorflow Lite Micro libraries.