A new algorithm for in‐band noise removal and HRV analysis in mouse ECG recordings (1169.7)

Fast and accurate analysis of ECG signals from awake and behaving mice can be challenging due to muscle noise, movement artifact, ECG morphology changes, and rapid changes in heart rate variability in response to experimental stimuli. A new patented technology, Multi‐Domain Signal Processing (MDSP)TM, has been shown to reduce in‐band noise in ECG recordings from non‐human primates by up to 21 dB without distorting ECG morphology (M. Brockway and R. Hamlin, Journal of Pharmacological and Toxicological Methods, 2011). In this study, we evaluated MDSP performance on ECG signals from active telemetered (ETA‐F20, DSI, St. Paul, MN) C57BL/6 mice with signal‐to‐noise ratio ranging from ‐9 to 30 dB (per ANSI/AAMI EC‐57 standard). In‐band noise removal was evaluated by contaminating a clean ECG strip with varying levels of white noise per ANSI/AAMI EC‐57 methodology. In‐band noise was reduced by up to 15 dB (82% amplitude reduction) with virtually no ECG waveform distortion. The QRS detection accuracy achieved by MDSP required screening of < 0.5% of beats with no need to preselect clean ECG segments prior to HRV analysis. HRV analysis was performed using traditional frequency domain and a novel time‐frequency domain wavelet‐based analysis and the discriminatory power of these two techniques was compared. Results show that MDSP demonstrates significant potential for efficient and accurate analysis of ECGs from freely moving mice, especially where it is important to discriminate rapid changes in HRV.