SignalProcessingCup
Discrete candidate analysis for heart rate monitoring using wrist-type photoplethysmographic signals during intensive exercise
Team
Kai Lin1, James Smith1, Ross Leon1, Robert Makepeace1, Nicholas Cummins1;2, Vidhyasaharan Sethu1
1: School of Elec. Eng. and Telecom., The University of New South Wales, Sydney Australia
2: ATP Research Lab, National ICT (NICTA), Australia
Abstract
Wrist-type photoplethysmographic (PPG) signals are an increasingly popular way to monitor heart rate during intensive exercise. However these signals are highly influenced by motion artefacts. This paper proposes a novel system for heart rate extraction termed Discrete Candidate Analysis (DCA). The DCA extracts a discrete set of possible heart rates from PPG signal, from which the sequence of most likely candidates is chosen based on accelerometric data, harmonic and temporal analysis. On a dataset of 23 PPG recordings, the proposed system obtained an average error of 3.02 beats per minute.
Fig. 1. DCA Overall system: sets of heart rate and artefact candidates are chosen from PPG and accelerometer signals respectively. They are analysed and one heart rate candidate is chosen for output
Report
Data
Software
Data
Images
Fig. 2. Comparison of Average Absolute Error (AAE) between TROIKA [5], system of [4] and proposed DCA system over 23 datasets