`
`A comparison of photoplethysmography and ECG recording to analyse heart
`rate variability in healthy subjects
`
`Article in Journal of Medical Engineering & Technology · November 2009
`
`DOI: 10.3109/03091900903150998 · Source: PubMed
`
`CITATIONS
`173
`
`4 authors, including:
`
`Guohua Lu
`Fourth Military Medical University
`
`39 PUBLICATIONS 820 CITATIONS
`
`SEE PROFILE
`
`Some of the authors of this publication are also working on these related projects:
`
`O Quam Gloriosum View project
`
`Deep Brain Stimulation View project
`
`READS
`2,330
`
`John Frederick Stein
`University of Oxford
`
`470 PUBLICATIONS 23,131 CITATIONS
`
`SEE PROFILE
`
`All content following this page was uploaded by John Frederick Stein on 28 July 2018.
`
`The user has requested enhancement of the downloaded file.
`
`APPLE 1015
`
`1
`
`
`
`Journal of Medical Engineering & Technology
`
`ISSN: 0309-1902 (Print) 1464-522X (Online) Journal homepage: http://www.tandfonline.com/loi/ijmt20
`
`A comparison of photoplethysmography and
`ECG recording to analyse heart rate variability in
`healthy subjects
`
`G. Lu, F. Yang, J. A. Taylor & J. F. Stein
`
`To cite this article: G. Lu, F. Yang, J. A. Taylor & J. F. Stein (2009) A comparison of
`photoplethysmography and ECG recording to analyse heart rate variability in healthy subjects,
`Journal of Medical Engineering & Technology, 33:8, 634-641, DOI: 10.3109/03091900903150998
`To link to this article: https://doi.org/10.3109/03091900903150998
`
`Published online: 15 Dec 2009.
`
`Submit your article to this journal
`
`Article views: 1268
`
`Citing articles: 54 View citing articles
`
`Full Terms & Conditions of access and use can be found at
`http://www.tandfonline.com/action/journalInformation?journalCode=ijmt20
`
`2
`
`
`
`Journal of Medical Engineering & Technology, Vol. 33, No. 8, November 2009, 634–641
`
`Innovation
`
`A comparison of photoplethysmography and ECG recording to
`analyse heart rate variability in healthy subjects
`
`G. LU{{, F. YANGx, J. A. TAYLOR{ and J. F. STEIN*{
`
`{Department of Physiology, Anatomy and Genetics, University of Oxford, OX1 3PT, UK
`{Department of Biomedical Engineering, Fourth Military Medical University, Xi’an, 710032, PR China
`xDepartment of Experimental Teaching Centre, Fourth Military Medical University, Xi’an, 710032,
`PR China
`
`Measures of heart rate variability (HRV) are widely used to assess autonomic nervous system
`(ANS) function. The signal from which they are derived requires accurate determination of
`the interval between successive heartbeats; it can be recorded via electrocardiography (ECG),
`which is both non-invasive and widely available. However, methodological problems
`inherent in the recording and analysis of ECG traces have motivated a search for alternatives.
`Photoplethysmography (PPG) constitutes another means of determining the timing of
`cardiac cycles via continuous monitoring of changes in blood volume in a portion of the
`peripheral microvasculature. This technique measures pulse waveforms, which in some
`instances may prove a practical basis for HRV analysis. We investigated the feasibility of
`using earlobe PPG to analyse HRV by applying the same analytic process to PPG and ECG
`recordings made simultaneously. Comparison of 5-minute recordings demonstrated a very
`high degree of correlation in the temporal and frequency domains and in nonlinear dynamic
`analyses between HRV measures derived from PPG and ECG. Our results confirm that PPG
`provides accurate interpulse intervals from which HRV measures can be accurately derived in
`healthy subjects under ideal conditions, suggesting this technique may prove a practical
`alternative to ECG for HRV analysis. This finding is of particular relevance to the care of
`patients suffering from peripheral hyperkinesia or tremor, which make fingertip PPG
`recording impractical, and following clinical interventions known to introduce electrical
`artefacts into the electrocardiogram.
`
`Keywords: Autonomic nervous system; Electrocardiography; Heart rate variability;
`Photoplethysmography
`
`1. Introduction
`
`a
`is
`(HRV)
`variability
`rate
`heart
`of
`Analysis
`powerful tool used to evaluate the regulation of cardiac
`activity by the autonomic nervous system (ANS). Since its
`inception HRV has been proven to index foetal distress [1],
`reveal diabetic neuropathy
`[2]
`and uncover ANS
`
`*Corresponding author. Email: john.stein@dpag.ox.ac.uk
`All authors contributed equally to this work.
`
`pathology [3]. Importantly, HRV has also been shown
`to predict the mode of death in chronic heart failure
`[4],
`raising the prospect
`that HRV may prove a
`valuable guide to clinical
`intervention in cardiovascular
`disease [5].
`Whilst HRV analysis has previously been restricted
`to research applications,
`the increasing availability of
`
`Journal of Medical Engineering & Technology
`ISSN 0309-1902 print/ISSN 1464-522X online ª 2009 Informa UK Ltd.
`http://www.informaworld.com/journals
`DOI: 10.3109/03091900903150998
`
`3
`
`
`
`Photoplethysmography and ECG for analysis of heart rate variability
`
`significant amounts of computational power is making
`the widespread clinical use of HRV analysis feasible. It
`is therefore opportune to consider the possibility of
`improving methods for acquisition of the physiological
`signal from which HRV measures are derived.
`HRV measures are established by analysis of the temporal
`relationship between successive heartbeats. Conventionally this
`signal is determined by electrocardiography (ECG); each R-
`wave in the electrocardiogram is caused by depolarization of
`the main mass of the ventricular myocardium. However in
`theory any discrete event in the cardiac cycle may be repeatedly
`measured to produce a record of successive heartbeats.
`Cyclical oscillations in blood flow, which drive volumetric
`and oxygenation changes in the peripheral microvasculature,
`are directly driven by left ventricular contractions. Photo-
`plethysmography (PPG) is an optical technique capable of
`recording these changes in the microvasculature of peripheral
`tissues [6–9]. Modern PPG uses a single optical sensor, with
`a near-infrared emitter and detector integrated into a
`probe, which can be readily placed on the forefinger or
`earlobe. The probe is mechanically robust, reusable and
`comfortable to wear.
`In contrast
`to PPG, ECG recording uses Ag/AgCl
`electrodes attached to specific anatomical positions. Clin-
`ical ECG recording commonly uses 12 leads for determina-
`tion of the complex temporal dynamics of each cardiac
`cycle. However, for the purpose of HRV analysis only three
`electrodes are necessary to detect successive R waves, in
`accordance with Einthoven [10].
`ECG recordings are, however, often imperfect. Common
`sources of noise are those generated by physiological
`processes, including electromyograph contamination, signal
`interference and respiration induced baseline drift, as well as
`those generated by non-physiological influences such as
`power line interference and electrode contact movement. In
`addition, morphological variations in the ECG waveform
`and the high degree of heterogeneity in the QRS complex
`often make it difficult to identify R waves, which may
`preclude the accurate determination of R-R intervals (RRI).
`Here we ask whether PPG may provide a more reliable
`means of deriving heart rate records than ECG.
`Early applications of PPG in the investigation of ANS
`function involved identification of known cardiovascular
`functional correlates of autonomic pathology [11,12]. The first
`investigations to establish the high degree of correlation
`between PPG and ECG measures of interbeat intervals did not
`examine their applicability to HRV analysis [13, 14]; they did
`however raise the prospect of that PPG might provide an
`effective substitute for ECG in such analyses.
`Bolanos et al. [15] carried out a small pilot study in healthy
`participants (n¼ 2) to evaluate the equivalence of PPG and
`ECG in analysing HRV. Whilst this investigation suggested
`that PPG may be as useful as ECG, it remained to be
`determined whether this could be replicated in a larger sample.
`Subsequently, Selvaraj and colleagues [16] investigated the
`
`635
`same question in a slightly larger sample (n¼ 10) of healthy
`participants and confirmed the findings of Balanos et al.
`Balanos et al. [15] do not report the site of their PPG
`recording; Selvaraj and colleagues [16] recorded from the
`fingertip. However PPG recordings made at this site are known
`to be highly vulnerable to motion artefact [17–20]; whilst the
`heart rate signal can often be recovered by band-pass fil-
`tering, blood oxygen saturation is frequently lost. As both
`measures are sought from a multifunctional PPG probe to
`be used in clinical evaluation, artefact must be minimized.
`Free movement ordinarily entails significantly more motion
`of the hands than of the head, making the earlobe a more
`suitable site for PPG recording than the fingertip. Also
`earlobe recording is desirable in movement disorders such
`as Parkinson’s disease because hyperkinesia and tremor is
`maximal
`in the fingers [21]. Likewise psychologists are
`becoming increasingly interested in the applicability of
`HRV measures to understand neural responses to cognitive
`stimuli. In most of these paradigms finger presses are used
`to respond, and these interfere with PPG recording.
`Bolanos et al. [15] used a sampling rate of 196 Hz, whilst
`Selvaraj et al. [16] used 1 kHz. Since one of the identified
`potential benefits of PPG is the possibility of using this method
`for ambulatory or home based recordings we also sought to
`ascertain whether a lower PPG sampling rate, which would
`demand more modest data storage capacity, could match the
`measures of HRV derived from the 200 Hz ECG sampling
`frequency deemed sufficient for HRV analysis [22].
`Thus we sought to confirm and expand on previous reports
`that PPG may provide a valuable alternative to ECG in the
`assessment of HRV, and asked whether the same signal
`analysis techniques can be successfully applied to both
`earlobe PPG and 3-lead ECG signals in healthy subjects.
`Here we show under controlled research conditions that
`ECG and PPG derived measures of ANS function are
`similar; we compared completely overlapping 5-minute PPG
`and ECG recordings to compute HRV in both time and
`frequency domains and using nonlinear dynamic indices.
`
`2. Methods
`
`2.1. Signal recording
`
`A total of 42 subjects gave informed consent to participate
`in this study (34 males, 21.1 + 3.4; eight females, 20.8 +
`2.3). A structured interview determined that all participants
`were in good health and none reported symptoms of
`autonomic or cardiovascular disorder.
`For ECG recording, disposable Ag/AgCl resting ECG
`electrodes (Red DotTM -2330; 3M Company, Minnesota,
`USA) were attached to the right wrist (‘Ground’), right
`forearm (‘Negative’) and left forearm (‘Positive’) to enable
`recording of the Lead I trace. Wires from the electrodes were
`attached to an ECG sensor (PS-2111; Feedback Instruments,
`East Sussex, UK).
`
`4
`
`
`
`636
`
`G. Lu et al.
`
`where x represents the RR intervals, y represents the PP
`intervals; N is the number of intervals, and mx and my
`represent the mean of the RRIs and PPIs. A value of close
`to 1 indicates a high direct correlation, whereas values close
`to –1 indicate an inverse relation, with values close to 0
`indicating little or no relation.
`
`For recording the PPG signal, a transmission ear-clip
`pulse sensor (PS-2105; Feedback Instruments) was attached
`to the left earlobe. The pulse and ECG sensors were both
`connected to a single USB link (PS-2001; Feedback
`Instruments), which was in turn directly connected to a
`desktop computer. In order to minimize motion artefact in
`the PPG signal the wire connecting the PPG probe and
`USB link was attached to each participant’s neck using
`surgical tape (MicroporeTM - 1530-0, 3M Company).
`ECG and PPG signals were sampled at a frequency of
`200 and 100 Hz respectively. Both ECG and PPG signals
`were simultaneously recorded for 7 minutes using the Data-
`Studio software package (1.9.7r8; Feedback Instruments).
`Subjects remained semi-recumbent throughout the record-
`ing period and were instructed to minimize their movement.
`
`2.2. Extraction of HRV and PPV signals
`
`An experienced researcher (GL) selected completely over-
`lapping 5-minute ECG and PPG segments with minimal
`artefact. Raw ECG and PPG signals were preprocessed
`using a FIR low-pass filter with cut-off frequency of 40 Hz
`and a notch filter to reduce high frequency and power line
`interference. HRV analyses were performed with purpose-
`written algorithms, using the MATLAB software package
`(MATLAB version 6.5; The MathWorks, Inc., Natic,
`Massachusetts, USA).
`For the ECG recordings, the extraction method incorpo-
`rated a peak detection algorithm that found the time of occur-
`rence of each QRS complex in the filtered ECG signal [23], and
`then the durations between successive peak locations were
`calculated to produce a time series of R–R intervals (RRIs).
`For the PPG recordings, a neighbouring peak searching
`method was used to derive the peak events from the
`amplitude of the filtered PPG signals [24] and then the
`intervals between the successive detected peaks (PPIs) were
`calculated. All of the RRI and PPI time series underwent
`an initial automated editing process before a careful
`manual editing was performed by visual inspection.
`
`2.3. Measuring the similarity between RRIs and PPIs
`
`To demonstrate the similarity of the HRV waveforms, two
`parameters were directly computed from the RRI and PPI
`signals, namely their cross correlation coefficient and the
`mean squared error between them.
`
`P
`2.3.1. The cross-correlation coefficient. The cross-correla-
`tion coefficient was calculated using equation (1).
`q
`ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
`P
`P
`¼1 ðxi mxÞðyi myÞ
`r ¼
`P
`P
`P
`¼1 ðxi mxÞ2
`¼1 ðyi myÞ2
`q
`ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
` P
`
`P
`P
`P
`¼1 xiyi ð
`¼1 xiÞð
`¼1 yiÞ=N
`i ð
`¼1 xiÞ2=N
`i ð
`¼1 yiÞ2=N
`¼1 y2
`¼1 x2
`ð1Þ
`
`;
`
`N i
`
`N i
`
`N i
`
`N i
`
`N i
`
`N i
`
`N i
`
`N i
`
`N i
`
`N i
`
`¼
`
`2.3.2. The mean squared error.The mean squared error
`(MSE) was defined as:
`
`vuut
`ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
`
`XN
`
`MSE ¼
`
`ðxi yiÞ2
`
`i¼1
`
`ð2Þ
`
`where x represents the R-R intervals, y represents the
`PP intervals; N is the number of intervals. A value of MSE
`close to 0 indicates the two waveforms were very similar.
`
`2.4. Measuring parameters in HRV and PPV recordings
`
`2.4.1. Time domain parameters. Four parameters were
`calculated from the time domain HRV and PPV recordings:
`the mean interpulse interval (mean NN), the standard
`deviation of the interpulse intervals (SDNN), the square
`root of the mean squared differences of successive inter-
`pulse intervals (RMSSD), and the proportion of differences
`of successive interpulse interval exceeding 50 ms, known as
`pNN50; this was derived by dividing n RR 450 by the
`total number of interpulse intervals [25].
`
`2.4.2. Frequency domain parameters. The RRI and PPI
`sequences were cubic interpolated and resampled at 4 Hz.
`Then low frequency (LF) power (0.04–0.15 Hz), high
`frequency (HF) power (0.15–0.4 Hz) and the ratio of LF
`to HF power were calculated in accordance with previously
`published standards for the spectral analysis of HRV [25],
`yielding three frequency domain measures. Power fre-
`quency (Hz) was converted to ms2 using the fast Fourier
`transform (FFT) employing 1024 points using in-house
`software.
`
`2.4.3. Poincare´ parameters. The Poincare´ plot is one of the
`most widely used techniques for nonlinear HRV analysis. It
`is a plot of each RR interval against the previous one.
`From a Poincare´ plot, two nonlinear parameters SD1 and
`SD2 can be calculated [26]:
`
`ð3Þ
`
`ð4Þ
`
`r
`ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
`Varðxnþ1 xxÞ
`r
`ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
`2SDNN2 1
`2
`
`SD2
`1
`
`1 2
`
`SD1 ¼
`
`SD2 ¼
`
`where x represents the PPI or RRI sequences, symbol Var
`is the variance of the differences in successive RRI or PPI,
`SDNN is the standard deviation of the RRIs or PPIs. SD1
`
`5
`
`
`
`Photoplethysmography and ECG for analysis of heart rate variability
`
`637
`
`represents short term beat-to-beat variability of the data,
`and SD2 is long term beat-to-beat variability.
`
`2.5. Statistical analysis
`
`To assess how similar HRV parameters derived from PPV
`compared with those derived from the ECG,
`linear
`
`regression analysis was performed. Before analysis, raw
`values of all variables were examined for deviations from
`normality by the Kolmogorov–Smirnov test. Statistical
`analysis was performed in SPSS@ (v13, SPSS Inc., Chicago,
`IL, USA) and graphs were plotted using Origin@ (v7.5776,
`Northampton, MA, USA).The level of significance was set
`at p 50.05 (two-tailed).
`
`Figure 1. Comparison of the HRV and PPV signals derived from ECG and PPG in a single representative subject: (a) RRIs
`derived from 5-minute ECG recording, (b) PPIs derived from 5-minute simultaneous PPG recording, (c) squared difference
`between RRIs and PPIs.
`
`Figure 2. HRV and PPV Poincare´ plots for a single representative subject: (a) Poincare´ plot of HRV derived from ECG,
`(b) Poincare´ plot of PPV derived from PPG. SD1 and SD2 index the short and long term HRV or PPV respectively.
`
`6
`
`
`
`638
`
`3. Results
`
`G. Lu et al.
`
`Figures 1(a) and 1(b) show 300-s segments of HRV and
`PPG signals from a single representative subject. The top
`
`two panels show that they are nearly identical with an
`overall correlation coefficient of 0.98 (p 50.05). Figure 1(c)
`shows the squared difference between HRV and PPV
`signals; this was tiny (mean squared difference ¼ 0.000069).
`
`Figure 3. Comparison of RRI and PPI sequence: (a) box plot of range of correlation coefficients in different subjects,
`(b) mean squared error.
`
`Figure 4. Time domain linear regression analysis RRI versus PPI: (a) average of RRIs or PPIs, (b) SDNN of RRIs or PPIs,
`(c) RMSSDD of RRIs or PPIs, (d) pRR50 of RRIs or PPIs.
`
`7
`
`
`
`Photoplethysmography and ECG for analysis of heart rate variability
`
`639
`
`Figure 5. Linear regression of parameters in frequency domain derived from HRV and PPV: (a) logarithm of the LF power,
`(b) logarithm of the HF power, (c) ratio of the LF to HF.
`
`Figure 6. Linear regression of SD1 and SD2 from Poincare´ plot derived from RRI and PPI: (a) SD1, (b) SD2.
`
`Figure 2 shows HRV and PPV Poincare´ plots derived from
`ECG and PPG. SD1 and SD2 index the short and long
`term HRV or PPV respectively.
`Figure 3 compares RRI and PPI results across subjects.
`Panel (a) shows strong correlations in the majority of
`subjects. The minimum value was 0.14, but the median was
`0.91,
`lower quartile 0.64, upper quartile 0.98 and the
`
`maximum value was 0.99. Panel (b) shows similar data for
`the rms differences. The minimum value was 0.000023, the
`median was 0.00029, lower quartile was 0.00009, upper
`quartile was 0.0024 and the maximum value was 0.035.
`Linear regression analyses of parameters derived from
`HRV and PPV in the time and frequency domains and the
`Poincare´ plot are shown figures 4–6. Figure 4(a) through
`
`8
`
`
`
`640
`
`G. Lu et al.
`
`(d) show significant correlation in mean NN, SDNN,
`RMSSD and pNN50 (r¼ 0.99, p 5 0.0001, n ¼ 42; r ¼ 0.99,
`p 5 0.0001,n ¼ 42; r ¼ 0.97, p 5 0.0001, n ¼ 42; r ¼ 0.95,
`p 5 0.0001, n ¼ 42). Figure 5 (a) to (c) show significant
`correlation in LF power, HF power and LF/HF ratio
`(r ¼ 0.99, p 5 0.0001; r ¼ 0.98, p 5 0.0001; n ¼ 42; r ¼ 0.97,
`p 5 0.0001). Figure 6 (a) and (b) show that significant
`correlation in SD1 and SD2 (r¼ 0.95, p 5 0.0001; n ¼ 42;
`r ¼ 0.99, p 5 0.0001, n ¼ 42).
`
`4. Discussion and conclusions
`
`Recent reports suggest that HRV measurement may prove
`an effective means of detecting early cardiovascular disease
`[27–29] as well as predicting the mode of death in chronic
`heart failure [4]. Consequently, HRV is attracting increas-
`ing interest
`from clinicians as a potentially valuable
`indicator for choosing prophylactic cardiovascular inter-
`ventions.
`researchers have argued that PPG
`A number of
`recording may prove more convenient than ECG for the
`measurement of HRV in clinical environments and for
`ambulatory recording [15, 16]. But many additional
`physiological parameters,
`including blood oxygenation
`and ventilatory rate, can be simultaneously derived from
`a single PPG recording. This raises the prospect that PPG
`recording could provide more clinically valuable informa-
`tion; this could be recorded by patients in their own home
`and transferred to medical professionals for offsite analysis
`and evaluation. Our results support this proposition by
`demonstrating that HRV analysis of signals derived from
`3-lead ECG and earlobe PPG recordings are almost
`identical; the pulse sensor is therefore a reliable means of
`recording a signal from which HRV measures can be
`derived, at least in healthy subjects at rest.
`ECG recording is often beset with electrical artefacts,
`such as those arising from deep brain or spinal cord
`stimulators [30–34]. In these situations PPG could prove
`superior for HRV analyses. In addition, signals derived
`from currently available PPG probes are susceptible to
`significant motion artefacts that are difficult to eliminate,
`and limit
`their current use in ambulatory recording.
`Integration of accelerometers into the PPG sensor enclo-
`sure could provide a means of quantifying the displacement
`of the sensor during recording [35]. Then signal analysis
`techniques could be used to reduce motion artefacts in the
`PPG signal, which might enable PPG to be effectively used
`for long-term recording in mobile patients.
`Oscillations in blood volume in the periphery are phase
`delayed in relation to the ECG [36]. In any given vascular
`bed this delay will depend on the elasticity of the vessels
`through which the blood passes from the heart, which in
`turn depends on their variable tone. These factors limit the
`capacity of PPG to accurately record the temporal
`dynamics of small numbers of cardiac cycles. However
`
`since pulse transit time shows minimal variation in healthy
`subjects at rest [37–39] and HRV analysis is dependent on
`recordings spanning many cardiac cycles, measures of HRV
`derived from PPG ought not to be significantly influenced
`by the degree of phase misalignment between the activity of
`the heart and the PPG signal when dynamic variation in
`vascular tone is minimal. This was supported by our data,
`where deviation from identity was seen to similar extents in
`Poincare´ plots of both RRI and PPI (figure 2). Therefore
`observed SD1 and SD2 indices exceeding zero are likely to
`reflect underlying variability in cardiac cycle durations
`rather than the influence of dynamic vascular tone on pulse
`transit time . It remains to be determined whether the close
`correlation between HRV measures derived from ECG and
`PPG will be robust in the face of variable conditions such
`as those encountered during ambulatory recordings.
`The potential advantages of PPG over ECG to derive
`HRV warrant further investigation in both ambulatory
`patients and those whose treatment is likely to generate
`electrical artefacts in their electrocardiogram.
`
`References
`
`[1] Hon, E.H. and Lee, S.T., 1963, Electronic evaluation of the fetal heart
`rate pattern preceding fetal death, further observation. American
`Journal of Obstetrics & Gynecology, 87, 814–826.
`[2] Ewing, D.J., Martyn, C.N., Young, R.J. and Clarke, B.F., 1985, The
`value of cardiovascular autonomic function tests: 10 years experience
`in diabetes. Diabetes Care, 8, 491–498.
`[3] Tulen, J.H.M., Man in ’t Veld, A.J., van Steenis, H.G. and Mechelse,
`K., 1991, Sleep patterns and blood pressure variability in patients with
`pure autonomic failure. Clinical Autonomic Research, 1, 309–315.
`[4] Guzzetti, S., Rovere, M.T.L., Pinna, G.D., Maestri, R., Borroni, E.,
`Porta, A., Mortara, A. and Malliani, A., 2005, Different spectral
`components of 24 h heart rate variability are related to different modes
`of death in chronic heart failure. European Heart Journal, 26, 357–362.
`[5] Chattipakorn, N., Incharoen, T., Kanlop, N. and Chattipakorn, S.,
`2007, Heart rate variability in myocardial infarction and heart failure.
`International Journal of Cardiology, 120, 289–296.
`[6] Challoner, A.V.J. and Ramsay, C.A., 1974, A photoelectric plethys-
`mograph for the measurement of cutaneous blood flow. Physics in
`Medicine and Biology, 19, 317–328.
`[7] Hertzman, A.B., 1937, Photoeletronic plethysmography of the fingers
`and toes in man. Proceedings of the Society for Experimental Biology
`and Medicine, 37, 529–534.
`[8] Jago, J.R. and Murray, A., 1988, Repeatability of peripheral pulse
`measurements on ears, fingers and toes using photoelectric plethys-
`mography. Clinical Physics and Physiological Measurement, 9,
`319–329.
`[9] Kamal, A.A., Harness, J.B., Irving, G. and Mearns, A.J., 1989, Skin
`photoplethysmography—a review. Computer Methods and Programs
`in Biomedicine, 28, 257–269.
`[10] Eithoven, W., 1903, Die galvanometrische registering des menschli-
`chen electrocardiograms, zugleich eine beurtheilung der anwedung des
`kapillar-electrometers in physiologie. European Journal of Physiology,
`99, 472–480.
`[11] Broome, I.J. and Mason, R.A., 1988, Identifcation of autonomic
`dysfunction with a pulse oximeter. Anaesthesia, 43(1010), 833–836.
`[12] Modi, K.D., Sharma, A.K., Mishra, S.K. and Mithal, A., 1997, Pulse
`oximetry for the assessment of autonomic neuropathy in diabetic
`patients. Journal of Diabetes and its Complications, 11, 35–39.
`
`9
`
`
`
`Photoplethysmography and ECG for analysis of heart rate variability
`
`641
`
`[13] Teng, X.F. and Zhang, Y.T., 2003, Study on the peak interval
`variability of photoplethysmographic signals. In Proceedings of the
`IEEE EMBS APBME, Kyoto-Osaka-Nara, Japan, 20–22 October,
`pp. 140–141.
`[14] Johnston, W. and Mendelson, Y., 2005, Extracting heart rate
`variability from a wearable reflectance pulse oximeter. In Proceedings
`of
`the IEEE 31st Annual Northeast Bioengineering Conference,
`Hoboken, NJ, USA, 2–3 April, pp. 157–158.
`[15] Bolanos, M., Nazeran, H. and Haltiwanger, E., 2006, Comparison of
`heart rate variability signal features derived from electrocardiography
`and photoplethysmography in healthy individuals. In Proceedings of the
`IEEE EMBS, New York, USA, 30 August–3 September, pp. 4289–4294.
`[16] Selvaraj, N., Jaryal, A., Santhosh, J., Deepak, K.K. and Anand, S.,
`2008, Assessment of heart rate variability derived from finger-tip
`photoplethysmography as compared to electrocardiography. Journal
`of Medical Engineering & Technology, 32, 479–484.
`[17] Barker, S.J. and Shah, N.K., 1997, The effects of motion on the
`performance of pulse oximeters in volunteers (revised publication).
`Anesthesiology, 86, 101–108.
`[18] Benoit, H., Costes, F., Feasson, L., Lacour, J.R., Roche, F., Denis, C.,
`Geyssant, A. and Barthelemy, J.C., 1997, Accuracy of pulse oximetry
`during intense exercise under severe hypoxic conditions. European
`Journal of Applied Physiology, 76, 260–263.
`[19] Norton, L.H., Squires, B., Craig, N.P., McLeay, G., McGrath, P. and
`Norton, K.I., 1992, Accuracy of pulse oximetry during exercise stress
`testing. International Journal of Sports Medicine, 13, 523–527.
`[20] Trivedi, N.S., Ghouri, A.F., Shah, N.K., Lai, E. and Barker, S.J.,
`1997, Effects of motion, ambient light, and hypoperfusion on pulse
`oximeter function. Journal of Clinical Anesthesia, 9, 179–183.
`[21] Uitti, R.J., Baba, Y., Wszolek, Z.K. and Putzke, D.J., 2005, Defining
`the Parkinson’s disease phenotype:
`initial symptoms and baseline
`characteristics in a clinical cohort. Parkinsonism & Related Disorders,
`11, 139–145.
`[22] Merri, M., Farden, D.C., Mottley, J.G. and Titlebaum, E.L., 1990,
`Sampling frequency of the electrocardiogram for spectral analysis of
`the heart rate variability. IEEE Transactions on Biomedical Engineer-
`ing, 37, 99–106.
`[23] So, H.H. and Chan, K.L, 1997, Development of QRS detection method
`for real-time ambulatorycardiac monitor, In Proceedings of the IEEE
`EMBS, Chicago, IL, USA, 30 October–2 November, pp. 289–292.
`[24] Chang, F.C., Chang, C.K., Chiu, C.C., Hsu, S.F. and Lin, Y.D., 2007,
`Variations of HRV analysis in different approaches. In Computers in
`Cardiology, Durham, NC, 30 September–3 October, pp. 17–20.
`[25] Malik, M., Bigger, J.T., Camm, A.J., Kleiger, R.E., Malliani, A.,
`Moss, A.J. and Schwartz, P.J., 1996, Heart rate variability: standards
`of measurement, physiological interpretation and clinical use. Eur-
`opean Heart Journal, 17, 354–381.
`
`[26] Brennan, M., Palaniswami, M. and Kamen, P., 2001, Do existing
`measures of Poincare plot geometry reflect nonlinear features of heart
`rate variability? IEEE Transactions on Biomedical Engineering, 48,
`1342–1347.
`[27] Hohnloser, S.H., Klingenheben, T., Zabel, M. and Li, Y.G., 1997,
`Heart
`rate variability as an arrhythmia risk stratifier after
`infarction. Pacing and Clinical Electrophysiology, 20,
`myocardial
`2594–2601.
`[28] Kudaiberdieva, G., Gorenek, B. and Timuralp, B., 2007, Heart rate
`variability as a predictor of sudden cardiac death. Anatolian Journal of
`Cardiology, 7, 68–70.
`[29] Sandercock, G.R. and Brodie, D.A., 2006, The role of heart rate
`variability in prognosis for different modes of death in chronic
`heart failure. Pacing and Clinical Electrophysiology, 29, 892–904.
`[30] Han, H., Kim, M.J. and Kim, J., 2007, Development of real-time
`motion artifact reduction algorithm for a wearable photoplethysmo-
`graphy. In Proceedings of the IEEE EMBS, 22–26 August, Lyon,
`France, pp. 1538–1541.
`[31] Constantoyannis, C., Heilbron, B. and Honey, C.R., 2004, Electro-
`cardiogram artifacts caused by deep brain stimulation. Canadian
`Journal of Neurological Sciences, 31, 293–294.
`[32] Frysinger, R.C., Quigg, M. and Elias, W.J., 2006, Bipolar deep brain
`stimulation permits routine EKG, EEG, and polysomnography.
`Neurology, 66, 268–270.
`[33] Knight, B.P., Michaud, G.F., Strickberger, S.A. and Morady, F.,
`1999, Electrocardiographic differentiation of atrial flutter
`from
`atrial fibrillation by physicians. Journal of Electrocardiology, 32,
`315–319.
`[34] Martin, W.A., Camenzind, E. and Burkhard, P.R., 2003, ECG artifact
`due to deep brain stimulation. Lancet, 361(9367), 1431.
`[35] Siddiqui, M.A. and Khan, I.A., 2003, Differential electrocardio-
`graphic artifact from implanted spinal cord stimulator. International
`Journal of Cardiology, 87, 307–309.
`[36] Allen, J. and Murray, A., 2002, Age-related changes in the peripheral
`pulse timing characteristics at the ears, fingers and toes. Journal of
`Human Hypertension, 16, 711–717.
`[37] Drinnan, M.J., Allen, J. and Murray, A., 2001, Relation between heart
`rate and pulse transit time during paced respiration. Physiological
`Measurement, 22, 425–432.
`[38] Johansson, A., Ahlstrom, C., Lanne, T. and Ask, P., 2006, Pulse wave
`transit time for monitoring respiration rate. Medical and Biological
`Engineering and Computing, 44, 471–478.
`[39] Porta, A., Gasperi, C., Nollo, G., Lucini, D., Pizzinelli, P., Antolini,
`R. and Pagani, M., 2006, Global versus local
`linear beat-to-beat
`analysis of the relationship between arterial pressure and pulse transit
`time during dynamic exercise. Medical and Biological Engineering and
`Computing, 44, 331–337.
`
`
`
`View publication statsView publication stats
`
`10
`
`