Developing a novel noise artifact detection algorithm for smartphone PPG signals: Preliminary results

Abstract

Pulsatile signals recorded from a smartphone are often corrupted with noise artifacts, which hampers accuracy of the peak detection and consequently leads to inaccurate heart rate estimation. In this paper, we propose a novel approach which uses an algorithm based on variable frequency complex demodulation (VFCDM) to detect noise artifacts in the smartphone's pulsatile signal recorded from a fingertip video. The ultimate goal is to increase the accuracy of atrial fibrillation (AF) detection. In the time-frequency spectra obtained from VFCDM, thresholds are imposed on both the magnitude of the dominant frequency component at each time instant and on the successive difference of the significant frequency component in the heart rate range to enable accurate noise artifact detection. For this preliminary analysis, the performance of the proposed method has been evaluated on 200 subjects; the data were collected during a smartphone-based AF screening study in India. The proposed method is shown to detect noise artifacts in pulsatile signals with 91.16% accuracy, demonstrating the potential to reduce false alarms when only data segments identified as clean are used for AF detection.