Noise Reduction Method based on Autocorrelation for Threshold-Based Heartbeat Detection

Detecting heartbeats is more challenging in an ambulatory condition due to a higher level of noise and artefacts compared to heartbeats recorded in a hospital setting. To reduce false detection and improve the performance of threshold-based beat detection algorithm, an autocorrelation method for noise reduction within noisy electrocardiogram (ECG) signals is presented in this paper. The proposed method contained two components: autocorrelation and refining process. The proposed work used the autocorrelation method to generate periods of heartbeats to refine and identify the correct QRS complex and remove the noise in the candidate QRS complex, thus reducing incorrect detections. Results from the experiment showed that the proposed method was capable of reducing the noise and improving the beat detection performance in noisy signals.