Tobón Vallejo, Diana Patricia
(2016).
Modulation spectrum analysis for noisy
electrocardiogram signal processing and applications.
Thèse.
Québec, Université du Québec, Institut national de la recherche scientifique, Doctorat en télécommunications, 157 p.
Résumé
Advances in wearable electrocardiogram (ECG) monitoring devices have allowed for new cardiovascular
applications to emerge beyond diagnostics, such as stress and fatigue detection,
athletic performance assessment, sleep disorder characterization, mood recognition, activity
surveillance, biometrics, and fitness tracking, to name a few. Such devices, however, are
prone to artifacts, particularly due to movement, thus hampering heart rate and heart rate
variability measurement and posing a serious threat to cardiac monitoring applications. To
address these issues, this thesis proposes the use of a spectro-temporal signal representation
called “modulation spectrum”, which is shown to accurately separate cardiac and noise components
from the ECG signals, thus opening doors for noise-robust ECG signal processing
tools and applications.
First, an innovative ECG quality index based on the modulation spectral signal representation
is proposed. The representation quantifies the rate-of-change of ECG spectral
components, which are shown to be different from the rate-of-change of typical ECG noise
sources. As such, a signal-to-noise ratio (SNR) like metric is proposed, termed modulation
spectral based quality index (MS-QI). Unlike existing quality metrics, MS-QI does not rely
on machine learning algorithms, can be performed on single-lead ECGs, and was shown to
perform accurately with synthetic ECGs, as well as ECGs recorded in real-world environments.
Based on insights obtained from the MS-QI metric, a new adaptive ECG enhancement
algorithm is then proposed based on the principle of bandpass filtering in the modulation
spectral domain. The algorithm was tested on synthetic and recorded (extremely noisy)
ECG databases. Experimental results show the proposed algorithm outperforming a stateof-
the-art wavelet-based enhancement algorithm in terms of heart rate (HR) error percentage
measurement, signal-to-noise ratio (SNR) improvement, and ECG kurtosis; the latter is a
widely-used ECG quality metric. These findings suggest that the proposed algorithm can be
used to enhance the quality of wearable ECG monitors even in extreme conditions, thus it
can play a key role in athletic peak performance training/monitoring.
Moreover, wearable ECG monitoring applications are burgeoning and typically rely on
estimates of heart rate variability (HRV). Such applications require small computational footprint
and cannot rely on enhancement and HRV analysis, thus a stand-alone HRV metric is
needed. HRV indices have been proposed based on time- and frequency-domain analyses of
the ECG, as well as via non-linear approaches. These methods, however, are very sensitive
to ECG artefacts, thus limiting the number of applications involving noisy ECGs (e.g., athletic peak performance training). Typically, ECG enhancement is performed prior to HRV
computation to overcome this limitation. Existing enhancement algorithms, however, are
not accurate in very noisy scenarios. Hence, an alternate approach is proposed based on
the modulation spectrum. By quantifying the rate-of-change of ECG spectral components
over time, we show that heart rate estimates can be reliably obtained even in extremely noisy
signals, thus bypassing the need for ECG enhancement. The so-called MD-HRV (modulation
domain HRV) is tested on synthetic and recorded noisy ECG signals and shown to outperform
several benchmark HRV metrics computed post-enhancement. These findings suggest that
the proposed MD-HRV metric is well-suited for ambulant cardiac monitoring applications,
particularly those involving intense movement.
Finally, a quality-aware ECG monitoring application is presented based on the proposed
MS-QI. Wearable ECG devices are increasingly being used in telehealth applications, particularly
for patient monitoring applications. Representative devices include watches, chest
straps, and even smart clothing via textile ECG sensors. Such lower-cost sensors, however,
are extremely sensitive to movement, thus pose a serious threat to such ECG streaming applications.
For example, transmission bandwidth, battery life, and/or storage space can be
spent with ECG segments that convey little cardiac information due to the high levels of
noise present. Moreover, noisy signals may cause false alarms in automated patient monitoring
systems, thus increasing the burden on medical personnel. Here, by employing the
MS-QI to discriminate usable from non-usable ECG segments, a quality-aware storage protocol
was implemented where storage of cardiac parameters was only performed on the usable
segments. When tested with a smart shirt under three conditions, namely sitting, walking
and running, the proposed quality-aware application resulted in storage savings of 65%.
Type de document: |
Thèse
Thèse
|
Directeur de mémoire/thèse: |
Falk, Tiago H. |
Co-directeurs de mémoire/thèse: |
Maier, Martin |
Mots-clés libres: |
denoising; electrocardiogram; heart rate; heart rate variability; modulation
spectrum; quality index; telehealth; wearables |
Centre: |
Centre Énergie Matériaux Télécommunications |
Date de dépôt: |
10 avr. 2019 15:17 |
Dernière modification: |
10 avr. 2019 15:17 |
URI: |
http://espace.inrs.ca/id/eprint/8029 |
Actions (Identification requise)
 |
Modifier la notice |