Comparison of wavelet transformation and adaptive filtering in restoring artefact-induced time-related measurement

Photoplethysmography (PPG) can be used in tin-re-related measurements such as heart rate (HR) and pulse transit time (PTT) estimations in the medical fields. The accuracy of these two parameters is heavily dependent on the minimal phase variability of the PPG signals. Moreover, motion artefact is a common phenomenon that can contaminate the PPG signals. This paper compares the capabilities of two signal processing techniques: digital adaptive filtering and discrete wavelet transformation, in restoring artefact-induced PPG signals during two regulated mild movements. HR comparison was evaluated against estimates attained from an electrocardiogram (ECG) while PTT evaluation was based on a reference PPG source free of artefacts. Comparison criteria was based on documented evidences that a HR difference of < 10 beats per minute (bpm) and relative PTT changes of < 15 ms can be considered as normal. The obtained results reveal that wavelet transformation is limited in restoring corrupted PPG signals for both HR and PTT measurements. It can also be observed its inherent algorithm can induce unwarranted phase variability that can compromise clinical interpretation of these two physiologic parameters. Promising signal processing algorithms may be attractive for its improved image and/or signal restoration however. its suitability in time-related measurements warrants further investigations. (c) 2006 Elsevier Ltd. All rights reserved.