Evaluating FPGA-based denoising techniques for improved signal quality in electrocardiograms
G. Keerthiga,S. Praveen Kumar
DOI: https://doi.org/10.1007/s10470-024-02277-w
IF: 1.321
2024-06-26
Analog Integrated Circuits and Signal Processing
Abstract:The alarming mortality rates associated with cardiac abnormalities emphasize the critical need for early and accurate detection of heart disorders to mitigate severe health consequences for patients. Electrocardiograms (ECG) are commonly employed instruments for the examination of cardiac disorders, with a preference for noise-free ECG signals to ensure precise interpretation. However, ECG signal recordings are susceptible to environmental interferences, including patient movement and electrode positioning. This paper introduces a hardware implementation for denoising ECG signals, leveraging a novel method by integrating high-order Synchrosqueezing Transform, Detrended Fluctuation Analysis, and Non-Local-Mean filter optimized by Particle Swarm Optimization (HSST-DFA-PSO-NLM) techniques on Field-Programmable Gate Array (FPGA) platforms. FPGA-based processing units are chosen for their outstanding performance attributes, including high re-programmability, speed, architectural flexibility, and low power consumption, resulting in efficient signal processing. The effectiveness of the designed filtering algorithm is evaluated using key criteria, including Signal-to-Noise Ratio (SNR) and Root Mean Square Error (RMSE) for performance assessment. Additionally, resource utilization metrics such as Look-Up Tables (LUTs), Flip Flops, and DSP Blocks, as well as power consumption measures including dynamic power and static or leakage power, are analysed across various FPGA boards (Virtex and Zedboards) utilizing the VIVADO environment. Comparative analyses are conducted to identify the most suitable FPGA board for implementation, highlighting the superior performance of the proposed design. Remarkably, the proposed denoising solution gives excellent SNR of 29.56, 29.68, and 28.86 by denoising various ECG noises. The RMSE attained by the model is also less than 0.05. This research advances the field of cardiac disorder detection by providing a reliable and efficient FPGA-based solution for ECG signal denoising, thereby enhancing the accuracy of early diagnosis and treatment.
engineering, electrical & electronic,computer science, hardware & architecture