Content uploaded by Tomas Ysehak Abay
Author content
All content in this area was uploaded by Tomas Ysehak Abay on Feb 17, 2019
Content may be subject to copyright.
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 62, NO. 9, SEPTEMBER 2015 2187
Reflectance Photoplethysmography as Noninvasive
Monitoring of Tissue Blood Perfusion
Tomas Ysehak Abay∗, Student Member, IEEE, and Panayiotis A. Kyriacou, Senior Member, IEEE
Abstract—In the last decades, photoplethysmography (PPG) has
been used as a noninvasive technique for monitoring arterial oxy-
gen saturation by pulse oximetry (PO), whereas near-infrared spec-
troscopy (NIRS) has been employed for monitoring tissue blood
perfusion. While NIRS offers more parameters to evaluate oxy-
gen delivery and consumption in deep tissues, PO only assesses
the state of oxygen delivery. For a broader assessment of blood
perfusion, this paper explores the utilization of dual-wavelength
PPG by using the pulsatile (ac) and continuous (dc) PPG for the
estimation of arterial oxygen saturation (SpO2) by conventional
PO. Additionally, the Beer–Lambert law is applied to the dc com-
ponents only for the estimation of changes in deoxyhemoglobin
(HHb), oxyhemoglobin (HbO2), and total hemoglobin (tHb) as in
NIRS. The system was evaluated on the forearm of 21 healthy
volunteers during induction of venous occlusion (VO) and total
occlusion (TO). A reflectance PPG probe and NIRS sensor were
applied above the brachioradialis, PO sensors were applied on
the fingers, and all the signals were acquired simultaneously. While
NIRS and forearm SpO2indicated VO, SpO2from the finger did
not exhibit any significant drop from baseline. During TO, all the
indexes indicated the change in blood perfusion. HHb, HbO2,and
tHb changes estimated by PPG presented high correlation with the
same parameters obtained by NIRS during VO (r2=0.960, r2=
0.821, and r2=0.974, respectively) and during TO (r2=0.988, r2
=0.940, and r2=0.938, respectively). The system demonstrated
the ability to extract valuable information from PPG signals for a
broader assessment of tissue blood perfusion.
Index Terms—Beer–Lambert law, near-infrared spec-
troscopy (NIRS), optical sensors, photoplethysmography (PPG),
physiological monitoring, pulse oximetry (PO).
I. INTRODUCTION
NONINVASIVE optical technologies have contributed sig-
nificantly in the continuous and noninvasive monitoring
of tissue blood perfusion. Several modalities, such as photo-
plethysmography (PPG), pulse oximetry (PO), laser Doppler
flowmetry, near-infrared spectroscopy (NIRS), and reflectance
spectrophotometry, have been used in research and clinical set-
tings for the quantitative and qualitative assessment of different
tissue perfusion parameters [1]–[7].
PO is a noninvasive optical technique, which utilizes light at
two different wavelengths for the estimation of arterial oxygen
saturation. Light is applied to tissue and the light attenuations
at red (R) and infrared (IR) wavelengths are filtered, processed,
Manuscript received November 10, 2014; revised February 7, 2015; accepted
March 18, 2015. Date of publication March 30, 2015; date of current version
August 18, 2015. This work was supported by the Barts and The London NHS
Trust under Barts Charity Grant 832/1716. Asterisk indicates corresponding
author.
∗T. Y. Abay is with the School of Mathematics, Computer Sciences and
Engineering, City University London, EC1V 0HB London, U.K. (e-mail:
Tomas.Ysehak-Abay.1@city.ac.uk).
P. A. Kyriacou is with the School of Mathematics, City University London.
Digital Object Identifier 10.1109/TBME.2015.2417863
and separated in pulsatile (ac) and continuous (dc) PPG [8]–
[10]. The ac component reflects the changes in pulsatile arterial
blood volume and it is synchronous with the cardiac cycle. The
dc component depends on the nature of the tissue interrogated
and represents the relatively constant absorption of skin, venous
blood, nonpulsatile arterial blood, and total blood volume in the
light path [9]–[13]. Whereas ac components have a clinical im-
portance for their synchrony with the cardiac cycle, dc compo-
nents provide equally valuable information, such as hyperemic
or hypoaemic states, temperature changes, sympathetic outflow,
venous volume fluctuations, and other regulatory mechanisms
[13]. AC and dc components, at the two different wavelengths,
compose a physiological signal also known as PPG [9]. The
ratio between ac and dc components [i.e., ratio of ratios, see (1)]
is directly correlated to arterial blood oxygen saturation (SpO2)
by empirical curves, permitting the continuous and noninvasive
monitoring of arterial oxygen saturation by PO [8]–[10], [12].
PO has found many applications in clinical and research set-
tings. The most adopted use is the estimation of SpO2.The
technique provides continuous and noninvasive indication of
SpO2, presenting a valuable alternative to intermittent arterial
blood sampling. PO is nowadays used in clinical settings for
quick identification of possible hypoxic states, such as respi-
ratory failures, asthma, and chronic obstructive pulmonary dis-
eases [8], [12]. PO is regularly used in emergency medicine,
surgery, and neonatal care [8], [9], and it is mandatory during
anesthesia [14].
The PPG trace, however, is not only used for the estimation
of SpO2by PO. Several studies have been undertaken to prove
the ability of the PPG waveform on providing additional infor-
mation. The PPG has been used for heart rate analysis, pulse
transit time estimation, blood pressure estimation, respiration
rate, vascular tone assessment, arterial and venous assessment,
tissue viability, vasomotor function, thermoregulation, and as a
perfusion indicator [13]–[19].
Although PO is a powerful clinical tool, it presents some lim-
itations that have to be considered carefully by researchers and
clinicians. Limitations such as anemia, skin pigmentation, nail
polish, low perfusion, light interference, and venous pulsations
might introduce errors [8]–[10], [20], [21]. Even though SpO2
and PPG waveform analysis provide a wide range of informa-
tion, the majority of research in this field focuses in PO, which
is a technique that does not indicate the correct blood perfusion
status [12], [20]. Moreover, we did not find clear evidence in the
literature on whether PO (or the sole analysis of SpO2and ac
PPG signals) could be considered as a reliable indicator of tissue
blood perfusion when perturbations such as venous occlusion
occur [12], [13], [22], [23].
0018-9294 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
2188 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 62, NO. 9, SEPTEMBER 2015
NIRS is another optical technique, which estimates the
concentrations of oxyhemoglobin (HbO2), deoxyhemoglobin
(HHb), and total hemoglobin (tHb) in deep tissues [24], [25].
Light in the near-infrared spectrum (700–1000 nm) is utilized
in NIRS due to the relative transparency of human tissues in this
spectrum, permitting a deeper penetration of light [25]. In NIRS,
the Beer–Lambert law is used to derive the hemoglobin concen-
trations from the attenuation of light at two or more wavelengths
[24]. An index representing the tissue hemoglobin saturation is
then estimated from the ratio of HbO2and tHb [24]. In the last
decades, NIRS has found a growing range of applications as a
tissue perfusion monitoring technique. The technique has been
applied to measure cerebral and muscle perfusion, oxygen con-
sumption, and blood flow in tissues. It also found applications
in neonatal and foetal monitoring, somatic and splanchnic per-
fusion, peripheral vascular diseases, trauma medicine, sepsis,
and plastic surgery [24]–[30].
NIRS and PO differ in the assessment of tissue blood per-
fusion. Even though conventional PO is a valuable technique
adopted as a standard of care in many clinical environments,
it only permits the quantification of arterial oxygen saturation.
This restricts its use as an instrument for the assessment of
global oxygen delivery to tissues. Therefore, PPG so far has
been mainly utilized for the estimation of arterial oxygen satu-
ration by PO, with limited attempts in exploring the capability
and potentials of the technique for a more complete assess-
ment of tissue perfusion [31]. Alternatively, NIRS, by measuring
changes in hemoglobin concentrations, provides a more inclu-
sive representation of tissue blood perfusion. The tissue oxy-
genation index (TOI) and hemoglobin concentration changes
provide an indication of other possible alterations of tissue per-
fusion, such as arterial blockage, venous congestion, or oxygen
delivery/consumption mismatches [28], [32].
This paper presents the implementation and application of a
PPG instrument that offers a more comprehensive assessment
of tissue blood perfusion by PPG. The system comprises a pro-
cessing system and a reflectance PPG probe, and it uses ac and
dc components for the estimation of arterial oxygen saturation
(SpO2) by conventional PO. Furthermore, dc components are
processed separately, as in NIRS, to estimate changes of con-
centrations in HbO2, HHb, and tHb using the Beer–Lambert
law. The proposed system has been evaluated in healthy vol-
unteers during induced vascular occlusions on the forearm. A
commercial pulse oximeter and a NIRS system have been used
simultaneously as references throughout the measurements.
II. METHODS
Fig. 1 shows the block diagram of the system proposed in this
study and it comprises:
1) PPG processing system (ZenPPG);
2) reflectance PPG probe;
3) data acquisition (DAQ).
The optical components [light-emitting diodes (LEDs) and
photodiode]apply and detect light to and from the tissue. Red
and Infrared light separation is achieved in the PPG process-
ing system by multiplexer/demultiplexer and controlled by the
Fig. 1. Block diagram of the proposed system.
microcontroller technology. Raw PPG signals are acquired by
DAQ system. Signals are further separated into ac and dc com-
ponents and SpO2is estimated from their ratio. DC components
are used to estimate changes in HHb, HbO2, and tHb. Each part
of the system will be described in the following sections.
A. PPG Processing System—ZenPPG
As commercial pulse oximeters present limitations on dis-
playing and offering raw PPG signals, we opted for the use of the
ZenPPG, a custom-made research PPG system. The ZenPPG is a
flexible, dual-channel, battery-operated PPG processing system
developed by the Biomedical Engineering Group at City Univer-
sity London [33], [34]. The instrument permits the acquisition
of raw, dual-wavelength (red and infrared), and dual-channel
PPG signals [33].
The ZenPPG is composed of a modular architecture of ex-
changeable modules: System bus, power supply, core board, cur-
rent source, transimpedance amplifier, and probe module [33].
Customized PPG sensors can be connected to the system via
DB9 connectors on the probe board. The current-source board
is responsible for providing two separate currents to the LEDs.
Currents in one channel are fully digitally controlled, whereas
regulated by trimmers in the second channel [33]. The inter-
mittent switching of light RED/IR is achieved by multiplexers,
whereas the separation of the detected light in the two distinctive
wavelengths is achieved by demultiplexers [33]. The multiplex-
ing/demultiplexing process is controlled in the core board by
microcontroller technology (Atmel ATtiny 2313-20SU, Atmel
Corp., USA). The ON/OFF switching of RED/IR light sources
is carried at a frequency of 1 kHz. Once the light has been emit-
ted to the tissues, the detected light from the photodiodes (i.e.,
current) is converted to voltages by transimpedance amplifiers,
before being demultiplexed [33]. ZenPPG also comprises a 64-
p-i-n NI connector (National Instruments Corporation, Austin,
TX, USA) for the acquisition of the signals via LabVIEW and
the digital control of the LEDs currents [33]. The system allows
the full control of light emission and the simultaneous acqui-
sition of raw PPG signals (ac and dc) from the two indepen-
dent channels. ZenPPG has been recently used in PPG studies
ABAY AND KYRIACOU: REFLECTANCE PHOTOPLETHYSMOGRAPHY AS NONINVASIVE MONITORING OF TISSUE BLOOD PERFUSION 2189
Fig. 2. Reflectance forearm PPG probe. (a) Mechanical drawing, probe ele-
ments, and section. (b) Final manufactured reflectance PPG probe.
carried by the Biomedical Engineering Group at City University
London [34]–[36].
B. Forearm Reflectance Probe
A reflectance PPG probe was developed and manufactured in
order to be connected to ZenPPG and to acquire raw PPG signals
from the forearm. Fig. 2(a) shows the mechanical drawing of the
reflectance PPG probe. It comprises of a printed circuit board
(PCB), two red LEDs (KP-2012SRC-PRV, Kingbright, Taiwan),
and two infrared LEDs (KP-2012 SF4C, Kingbright, Taiwan)
with peak emission wavelengths at 660 and 880 nm, respec-
tively. These two wavelengths are opposite with respect to the
isobestic point and they have different hemoglobin absorption
coefficients, consenting the acquisition of R/IR signals for the
estimation of SpO2[8]. The LEDs were driven to a correspond-
ing radiant power between ∼1.5 and ∼3.6 mW/sr. A silicon pho-
todiode with a large active area of 7.5 mm2(TEMD5010X01,
Vishay Intertechnology Inc., USA) was selected for the acquisi-
tion of the backscattered light. The LEDs and photodiode were
placed at a center-to-center distance of 5 mm. This distance has
been considered in the literature as satisfactory for the acquisi-
tion of PPG signals with adequate signal-to-noise-ratio (SNR)
[37]–[39].
In order to guarantee the mechanical stability and to pro-
tect the optical components from ambient light interference, the
PCB was enclosed in a black case. The case was designed on a
3-D CAD design software and it was manufactured in polylac-
tic acid plastic by a 3-D printing technology. Black rubber was
used to shield the photodiode from light shunting and a layer
of clear epoxy medical adhesive (Dymax, 141-M) was used to
cover the LEDs and the photodiode. Fig. 2(b) shows the final
manufactured version of the reflectance PPG probe.
An identical PCB probe has been manufactured and enclosed
in a pulse oximeter clip for the acquisition of raw reflectance
PPG signals from the finger.
C. SpO2, HHb, HbO2, and tHb Estimation
As mentioned above, ac and dc PPG signals are used to es-
timate arterial oxygen saturation. The ratio of the components
(i.e. ratio of ratios, R), at the two wavelengths, is defined as
R=ACRED/DCRED
ACIRED/DCIRED
(1)
where ACRED and ACIRED are, respectively, the peak-to-
peak amplitudes of the pulsating arterial signal at red and
infrared wavelengths and DCRED and DCIRED are the rela-
tively constant values of light attenuation at the two respective
wavelengths.
The following standard empirical equation correlates the ratio
of ratios to arterial oxygen saturation, SpO2:
SpO2= 110 −25(R).(2)
The formula has been extensively used in PO, and it corre-
lates directly the ratio of ratios to arterial oxygen saturation [9].
Equations (1) and (2) were used in this study to estimate SpO2
from PPG signals.
DC components represent the amount of light reaching the
photodiode, thus, directly relating to the amount of blood vol-
ume interrogated by the light beam. Optical densities (OD), at
the two light wavelengths employed, were used in this study to
express changes in light attenuation due to variations in blood
volumes. ODs (or light attenuation changes) were calculated
from time changes of dc components as
ODλ1=ΔAλ1=lnDC (0)λ1
DCλ1(3)
ODλ2=ΔAλ2=lnDC(0)λ2
DCλ2(4)
where ΔAλ1and ΔAλ2are, respectively, the changes in light
attenuation at the wavelengths λ1and λ2, ln is the natural loga-
rithm, DC(0)λ1and DC(0)λ2are the dc values at the beginning
of the measurement for both wavelengths, and DCλ1and DCλ2
are the dc values throughout the entire measurement.
The modified Beer–Lambert law is extensively used in NIRS
and it correlates the attenuation Aof a monochromatic light
beam to the absolute values of chromophore concentration,
knowing the scattering contribution in the light path [24].
Equation (5) shows the generic form of the modified Beer–
Lambert law
Aλ=lnI0
I=ελ·[C]·d·DPF + G(5)
where Aλis the light attenuation at the wavelength λ,Iis the
light intensity detected, I0is the light emitted, [C]is the con-
centration of the chromophore C,ελis the extinction coefficient
of Cat wavelength λ,dis the distance between light emitter and
photodetector, DPF is the differential pathlength factor, and G
is a scattering factor. Continuous wave instruments are unable
to determine the factor G[24], [25], thus limiting to the only
estimation of relative changes in hemoglobin concentrations.
As our system is a continuous wave instrument (i.e., PPG sys-
tem), we opted for a differential approach, permitting only the
2190 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 62, NO. 9, SEPTEMBER 2015
calculation of relative changes in hemoglobin concentrations
from the start of the measurement. This approach permits to as-
sume that scattering in the tissue is constant during the measure-
ment, thus eliminating Gfrom (5) [24], [25]. Equations (6) and
(7) show the system of linear equations for a dual-wavelength
differential approach used in our PPG system
ΔA660 =εHHb660 Δ[HHb] + εHbO2660 Δ[HbO2]·d·DPF (6)
ΔA880 =εHHb880 Δ[HHb] + εHbO2880 Δ[HbO2]·d·DPF (7)
where ΔA660 and Δ880 are the attenuations at 660 and 880 nm,
εHHb and εHbO2 are, respectively, the extinction coefficients of
HHb and HbO2at the two wavelengths, Δ[HHb]and Δ[HbO2]
are the changes in concentration of the two hemoglobins, dis
the distance between the light emitter and photodetector, and
DPF is the differential pathlength factor.
The system of linear equations composed by (6) and (7) yields
the solutions for the changes in hemoglobin concentrations
Δ [HHb] = ΔA880εHbO2660 −ΔA660 εHbO2880
εHbO2660 εHHb880 −εHbO2880 εHHb660·d·DPF
(8)
Δ[HbO
2]= ΔA660εHHb880 −ΔA880 εHHb660
εHbO2660 εHHb880 −εHbO2880 εHHb660·d·DPF
(9)
Δ [tHb] = Δ [HHb] + Δ[HbO2].(10)
However, DPF values for the distance adopted in our PPG
system (i.e., 5 mm) are not available in the literature. As a full
estimation of the optical pathlength in the tissue was not fea-
sible, we opted for representing the results as relative changes
of hemoglobins concentrations in mM·cm ([absolute concentra-
tion]×[optical pathlength]).
D. NIRS System
A commercial NIRS system (NIRO 200NX, Hamamatsu Pho-
tonics K. K., Japan) was used as a reference for the estimation
of relative changes in concentration of HHb, HbO2, and tHb.
This NIRS device uses LEDs for the emission of light at three
different wavelengths: 735, 810, and 850 nm [40]. Two silicon
photodiodes, at a small spacing between each other, detect the
backscattered light and spatially resolved spectroscopy is ap-
plied for the absolute estimation of hemoglobins concentrations
[41]. TOI is calculated as the ratio between HbO2and tHb as
TOI = HbO2
HbO2+ HHb.(11)
An emitter-photodetector spacing of 4 cm has been adopted
in this investigation, guaranteeing the deep penetration of the
light beam in tissues [24], [25].
E. Experimental Setup and Protocol
The system was evaluated on healthy volunteers by inducing
vascular occlusions on the forearm. Ethical approval was gained
from the Senate Research Ethics Committee at City University
London. Twenty-one (21) healthy volunteers (13 males and 8
females, mean age ±SD: 31.18 ±7.55) were recruited for the
investigation. Subjects with a history of cardiovascular disorders
were excluded from the study. The physiological measurements
were performed in the Biomedical Engineering Research Group
laboratories at a room temperature of 22 ±1°C. The volunteers
were seated in a comfortable chair and their left arm was rested
on a pillow to minimize vascular compression. The reflectance
PPG probe was positioned on the volar side of the left brachiora-
dialis, while the NIRS probe was placed above the same muscle,
proximal to the PPG probe. The probes were kept in place on the
skin by double-sided clear medical tape. In order to avoid even-
tual optical impediment, the clear medical tape was previously
cut at the light emitters and photodiodes locations. Care was
taken in not compressing the probes on the skin surface during
placement of the sensors and during the entire protocol. The
custom-made finger PPG probe was connected to the second
digit of the left hand, while a commercial transmittance pulse
oximeter sensor (Radical 7, Masimo Corp., USA) was placed on
the third digit of the same hand. A cuff pressure was positioned
on the left upper arm in order to induce vascular occlusions. The
cuff was connected to a sphygmomanometer for the full control
of the occlusions pressure. ECG was monitored and acquired
throughout the measurements.
The volunteers’ blood pressure was measured prior to the
commencing of the measurements. After recording 5 min base-
line measurements, the cuff was rapidly inflated to 60 mmHg
(inflation time <4–5 s) in order to induce venous occlusion and
maintained at this pressure for 2 min. The occlusion pressure
was then quickly released for 2 min (deflation time <1–2 s)
before reinflating the cuff for further 2 min. Total occlusion was
performed by inflating the cuff to 20 mmHg greater than the
volunteer’s systolic pressure. After 2 min of total occlusion, the
cuff pressure was finally released and DAQ was stopped as soon
as all the signals returned to baseline values.
F. Signal Acquisition, Processing, and Analysis
All the signals (raw PPGs, NIRS parameters, and pulse
oximeter signals) were digitized by two NI-PCIe6321 DAQ
cards and acquired on a LabVIEW virtual instrument (VI) (Na-
tional Instruments, USA). The analogue signals were acquired
at a sampling frequency of 400 Hz. The VI was developed
to acquire, filter, and display the measurements in real time,
while raw signals were directly saved in a text file for further
postacquisition analysis. Algorithms were used for the real-time
estimation and display of SpO2as in (1) and (2).
Offline analysis of the signals was performed on MAT-
LABR2013a (The Mathworks Inc., USA). AC PPG signals
were extracted by applying a bandpass digital filter (cutoff fre-
quencies: 0.5–4 Hz) to the raw PPG signals. DC components
were obtained with a low-pass digital filter (cutoff frequency:
0.1 Hz). SpO2was calculated by applying (1) and (2) toa3s
rolling window. Hemoglobin concentration changes and TOI,
along with SpO2and single-wavelength PPG from the commer-
cial pulse oximeter, were directly acquired from the devices’
analogue outputs. Hemoglobin changes were estimated from dc
PPG components by applying (6)–(11).
ABAY AND KYRIACOU: REFLECTANCE PHOTOPLETHYSMOGRAPHY AS NONINVASIVE MONITORING OF TISSUE BLOOD PERFUSION 2191
Fig. 3. Infrared PPG traces from the forearm and fingers. (a) ZenPPG re-
flectance PPG from the forearm. (b) ZenPPG reflectance PPG from the finger.
(c) Finger PPG from the commercial pulse oximeter.
TAB LE I
MEAN FOREARM AND FINGERS SPO2AND TOI VALUES DURING BASELINE,
VENOUS OCCLUSION,AND TOTAL OCCLUSION
Baseline Venous Occlusion Total Occlusion
Mean (%) SD (%) Mean (%) SD (%) Mean (%) SD (%)
SpO2Finger (ZenPPG) 98.3 2.0 98.9 2.6 78.3 1.8
SpO2Finger (Radical 7) 98.5 0.8 98.1 1.1 39.2 32.0
SpO2Arm (ZenPPG) 92.0 3.1 88.9 3.6 73.7 12.8
TOI (NIRO 200NX) 69.1 4.7 65.8 4.4 62.3 6.1
All the parameters were expressed as their mean (±SD). Mean
SpO2and TOI values were calculated during 1-min baseline seg-
ment, venous occlusion, and total occlusion. Spearman’s corre-
lation analysis was performed to determine correlation between
changes in concentration of hemoglobins estimated from dc PPG
components and NIRS during vascular occlusions. Mean incre-
ment (or drops) in hemoglobin concentrations were calculated
as means of differences between hemoglobin values at the be-
ginning and end of the occlusions. A Wilcoxon signed-rank test
was also used to determine any statistical significant difference
in the forearm SpO2from baseline to venous occlusion.
III. RESULTS
A. SpO2and TOI
Fig. 3 shows typical PPG traces acquired from the forearm
and fingers of one volunteer during baseline measurements. AC
PPGs from the forearm presented lower amplitudes compared
with the fingers.
SpO2was calculated as in (1) and (2) from the PPG signals
from both forearm reflectance probe and from the finger. Table I
shows the mean SpO2, TOI values, and SD for the forearm and
fingers during baseline, venous occlusion, and total occlusion.
SpO2estimated from the forearm during baseline mea-
surements presented a lower mean value (92.0±3.1%) when
Fig. 4. Mean and SD of SpO2and TOI for all the volunteers during venous
occlusion. Vertical lines represent inflation and deflation of the cuff. (a) Finger
SpO2from the commercial pulse oximeter. (b) Finger SpO2from ZenPPG.
(c) Forearm SpO2from ZenPPG. (d) Forearm TOI from NIRS.
compared with SpO2values acquired from the custom-made
finger probe and the commercial probe (98.3±2.0% and
98.5±0.8%, respectively). SpO2from the fingers did not
exhibit any considerable mean desaturation from baseline dur-
ing venous occlusion, while, in contrast, SpO2from the fore-
arm showed a statistically significant desaturation (Z=−2.72,
p=0.006) with a mean drop of 3.1±3.6%. TOI exhibited a
mean drop of 3.3±4.4% throughout venous occlusion. During
total occlusion, all the SpO2values have dropped as expected.
These values are erroneous and are not reliable readings orig-
inated from the absence of arterial pulsations and consequent
increase in R. The disruption of oxygen delivery during total
occlusion caused a decrease in HbO2concentration and a con-
sequent drop in TOI value. Fig. 4 shows the means and SD of
SpO2values and TOI during venous occlusion.
B. Hemoglobin Concentrations
Figs. 5–7 show the mean changes of HHb, HbO2, and tHb
during venous and total occlusions. The hemoglobin concentra-
tions changes estimated from dc PPG components followed the
same trend with the signals produced by the NIRS system. HHb
exhibited increase in its concentration in both occlusions. These
were due to stagnation of venous blood caused by the blockage
of its return in both occlusions. HHb concentrations estimated
by NIRS and ZenPPG presented a significant correlation during
both venous occlusion (r2=0.960,p<0.05) and total occlu-
sion (r2=0.988,p<0.05). At the end of both occlusions, the
concentrations of HHb returned to baseline value.
HbO2presented two different behaviors during the vascular
occlusions. During venous occlusion, the oxygen delivery to
2192 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 62, NO. 9, SEPTEMBER 2015
Fig. 5. Mean and SD of HHb changes during venous and total occlusions
from NIRS and ZenPPG. Left subplots: venous occlusion. Right subplots: total
occlusion. (a) NIRS HHb during venous occlusion. (b) NIRS HHb during total
occlusion. (c) ZenPPG HHb during venous occlusion. (d) ZenPPG HHb during
total occlusion. Vertical lines represent inflation and deflation of the cuff. Solid
lines: mean; dashed lines: SD.
Fig. 6. Mean and SD of HbO2changes during venous and total occlusions
from NIRS and ZenPPG. Left subplots: venous occlusion. Right subplots: total
occlusion. (a) NIRS HbO2during venous occlusion. (b) NIRS HbO2during
total occlusion. (c) ZenPPG HbO2during venous occlusion. (d) ZenPPG HbO2
during total occlusion. Vertical lines represent inflation and deflation of the cuff.
Solid lines: mean; dashed lines: SD.
tissues was not compromised; hence, an initial increase was
observed, followed by a steady state. This increase is believed
to be due to the impaired washout of blood caused by venous
occlusion. During total occlusion, the oxygen delivery to tissues
was impeded causing a constant drop of HbO2throughout the
occlusion. HbO2estimated by NIRS and ZenPPG presented
a high correlation during venous occlusion (r2=0.821,p<
0.05) and total occlusion (r2=0.940,p<0.05). At the release
of total occlusion, HbO2manifested an overshot caused by the
postischemic hyperemic response of the tissue.
Fig. 7. Mean and SD of tHb changes during venous and total occlusions
from NIRS and ZenPPG. Left subplots: venous occlusion. Right subplots: total
occlusion. (a) NIRS tHb during venous occlusion. (b) NIRS tHb during total
occlusion. (c) ZenPPG tHb during venous occlusion. (d) ZenPPG tHb during
total occlusion. Vertical lines represent inflation and deflation of the cuff. Solid
lines: mean; dashed lines: SD.
TAB LE I I
MEAN CHANGES (±SD) IN HHB,HBO2,AND THBCONCENTRATIONS (IN
MM·CM)DURING VENOUS AND TOTAL OCCLUSION
Venous Occlusion Total Occlusion
NIRS ZenPPG NIRS ZenPPG
HHb 0.257 (±0.072) 0.102 (±0.042) 0.209 (±0.055) 0.071 (±0.027)
HbO20.128 (±0.080) 0.115 (±0.063) −0.086 (±0.058) −0.026 (±0.044)
tHb 0.385 (±0.139) 0.2179 (±0.088) 0.122 (±0.046) 0.044 (±0.043)
Total hemoglobin was estimated by summing HHb and HbO2
as in (10). Thus, its increase during venous occlusion is mainly
due to the rise of HHb concentration. Total hemoglobin can be
used as an estimate of total blood volume in tissues; hence, its
increase during total occlusion represented the gradual accu-
mulation of blood volume due to venous blood stagnation and
steady arterial blood in-flow. During total occlusion, tHb showed
an initial increase, followed by a relatively constant trend.
Total hemoglobin estimated by NIRS and ZenPPG showed
high correlation during both venous occlusion (r2=0.974,
p<0.05) and total occlusion (r2=0.938,p<0.05). At the
release of venous occlusion, tHb returned to its baseline values.
An overshot of tHb was observed in all cases following the re-
lease of total occlusion, indicating the postischemic hyperemic
response.
Table II shows the mean increments and decrements of HHb,
HbO2, and tHb estimated by NIRS and ZenPPG during both
occlusions. The values were calculated as the means of the
differences in amplitude at the beginning and end of the occlu-
sion periods.
ABAY AND KYRIACOU: REFLECTANCE PHOTOPLETHYSMOGRAPHY AS NONINVASIVE MONITORING OF TISSUE BLOOD PERFUSION 2193
IV. DISCUSSION
Mean SpO2and TOI values during baseline measurements
were in their physiological ranges and they were in good agree-
ment with previous studies [23], [29], [42]. However, baseline
SpO2, which was estimated by the reflectance probe on the fore-
arm, presented lower values compared with SpO2calculated
from the fingertips. This has not to be considered as a physio-
logical difference in arterial blood saturation as both sensors es-
timated SpO2from the same arterial branch. Nevertheless, these
lower values for the forearm are in agreement with the trends
showed in an earlier similar study by Mendelson and McGinn
in which a reflectance PPG sensor was developed for the es-
timation of SpO2from the forearm and calf of healthy adults.
Regression analysis revealed lower SpO2values estimated from
the forearm and the calf when compared with a transmission
finger probe [43]. These differences were attributed to the dif-
ficulty to extract high amplitude PPG from locations with a
low capillary density, such as the forearm and calf [43]. In or-
der to facilitate the acquisition of PPG signals from the two
locations, Mendelson and McGinn incorporated a heating ele-
ment into the reflectance sensor to increase the skin temperature
[43]. However, in this study, we decided not to use the same
approach as heating the skin might affect the blood perfusion,
thus conflicting with vascular occlusions adopted in this study.
We also observed from our experience that low values of SpO2
on the forearm might be due to higher R-values caused by a
higher absorption of red light in the forearm when compared
with fingers. Moreover, the presence of large superficial veins
on the forearm may affect the accuracy of the SpO2values
calculated [44].
Except the forearm, all the estimated SpO2values did not
significantly vary during venous occlusion. In contrast, and as it
was expected, they dropped during total occlusion. The absence
of any relevant drop during venous occlusion should be consid-
ered as a serious limitation of PO on monitoring tissue blood
perfusion. Shafique et al. observed that significant drops in ac
PPG amplitude were seen only for occlusion pressures exceed-
ing the diastolic pressure (e.g., 75 mmHg). We think that this
inability of PO may introduce risks in situations when a more
complete perfusion monitoring is required. As PO relies on pul-
sating arterial blood for the estimation of SpO2, the absence of
pulsations in total occlusion causes the pulse oximeter to fail in
the calculation of SpO2[23]. Even though a desaturation during
venous occlusion was detected, the presence of arterial pulsa-
tions (i.e., ac PPG) and an unaffected SpO2on the fingers might
still be misjudged as an adequate blood perfusion [13].
TOIs measured from the forearm exhibited the expected drops
in agreement with previous studies when NIRS was used in
conjunction with vascular occlusions [29], [30]. During both
occlusions, TOI showed desaturations providing a fair indica-
tion of changes in tissue blood perfusion. Throughout venous
occlusion, the continuous increase in HHb concentration and
relative steady state of HbO2caused a constant fall in TOI in
all volunteers. The increase in HHb and the simultaneous de-
crease of HbO2during total occlusion caused a more severe TOI
desaturation.
SpO2and TOI are extensively used in clinical and research
settings [8]–[10], [12], [20]–[23], [28]–[30]. The SpO2only
provides oxygen delivery information (i.e., arterial blood oxy-
gen saturation) and it cannot be used as a sole indicator of tissue
blood perfusion. Its dependence on pulsatile arterial PPG com-
ponents restricts its use in situations where changes in perfusion
are not directly reflected in arterial blood. On the other side, TOI
(also called in the literature as StO2or rSO2) is not dependent on
arterial pulsations and its lower values, compared with SpO2,are
more representative of mixed blood oxygen saturation (arteries,
capillaries, and veins) [25]. Moreover, the index is directly cal-
culated from the ratio of HbO2and tHb (HbO2+HHb) and its
changes indicate perfusion variations directly affecting this ra-
tio. Thus, we believe that the estimation of ΔHHb, ΔHbO2,
and ΔtHb, along with SpO2, from PPG signals could im-
prove the ability of the technique in following changes in blood
perfusion.
The changes in HHb, HbO2, and tHb obtained from dc
PPG components were in agreement with the same parame-
ters measured by NIRS as can be seen in Figs. 5–7. The three
hemoglobins measured by NIRS followed the same trends de-
scribed previously in the literature when venous or total oc-
clusion occur [29], [32], [45], [46]. The hemoglobins changes
allowed the distinction of venous occlusion where it was unno-
ticed by conventional PO. At the release of occlusion, all the
hemoglobins returned to their baseline. In particular, HbO2and
tHb exhibited correctly the postischemic hyperemic response of
the tissue after the release of total occlusion. The initial increase
of tHb in total occlusion was due to the noninstantaneous infla-
tion of the cuff over systolic pressure and the consequent initial
build-up of venous blood. These results suggest that hemoglobin
concentration changes can be extracted from PPG signals. Very
recently, Akl et al. reported results from an animal study in
which PPG signals were used for monitoring liver tissue. In
their work, a PPG probe was placed on the parenchymal tissue
of two swine and NIRS principles and multilinear regression
analysis were performed to track changes in perfusion and oxy-
genation [47]. The quantitative differences observed in our study
between the PPG-derived and NIRS-derived hemoglobins have
to be related to the different light penetration depths. While the
large LEDs-photodiode spacing in NIRS allows the deep pene-
tration of light in tissues (approximately up to 2 cm) [25], [26],
the small separation distance in our reflectance PPG probe (i.e.,
5 mm) limits the penetration depth to only few millimeters in
the skin layers. The spacing employed in our reflectance PPG
probe has been chosen for the best acquisition of PPG signals
and as a tradeoff between the light intensities and photodetection
system. Another parameter affecting the light penetration depth
is the specific light wavelengths adopted to illuminate the tis-
sues; the NIRS monitor adopted in this study employs LEDs at
nominal wavelengths of 735, 810, and 850 nm, in contrast with
the 660/880 nm pair used in our PPG reflectance probe. Thus,
the shallower light penetration of our reflectance, PPG probe
could be considered as the cause for lower SpO2values esti-
mated from the forearm. The PPG signal detected by our probe
is likely to originate from capillaries structures within the skin.
As capillaries contain a mixture of arterial and venous blood,
2194 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 62, NO. 9, SEPTEMBER 2015
the calculated SpO2values have lower values compared with
fingers. Furthermore, the decrease of SpO2from the forearm
during venous occlusion may be caused by the shifting of arte-
rial/venous blood ratio in capillaries due to venous congestion.
However, superficial layers (skin) and deep tissue (brachiora-
dialis) share the same major arteries and veins; hence, occlu-
sions of these vessels produce changes in blood perfusion in
both compartments. Furthermore, monitoring skin circulation,
which serves as a blood reservoir and has a weaker autoreg-
ulation, can be used as an early indicator of changes in blood
perfusion in more internal organs [2], [48]. Although myoglobin
and hemoglobin in muscles cannot be distinguished when mea-
sured by NIRS [26], we decided to express the NIRS parameters
as “only hemoglobin” changes.
The rate of increase in HHb, or the decrease rate of HbO2,
measured by NIRS during venous or total occlusions have been
used as indicators of oxygen consumption [27], [28], [49].
Therefore, the values in Table II may be used as comparative
indicators of oxygen consumption. In particular, the part of tis-
sue interrogated by our PPG probe (i.e., skin) appears to have a
lower consumption rate when compared with tissues measured
by NIRS (i.e., muscle). These differences in metabolic rate are
in agreement with physiology tables on oxygen consumption of
various human body tissues [50].
Future work will focus on the improvement of the PPG
reflectance probe, for instance, the placement of the LEDs,
in order to achieve a more homogeneous interrogation of
volume of tissue. In addition, the probe could be further
miniaturized.
V. CONCLUSION
In this paper, we explored the capabilities of a PPG sys-
tem in providing additional information, when compared with
conventional PPG, on changes in blood perfusion during vas-
cular occlusions. The system estimated arterial blood oxygen
saturation as a conventional pulse oximeter and it applied the
Beer–Lambert law for the estimation of changes of concentra-
tion of HHb, HbO2, and tHb. The evaluation of the system in
healthy volunteers demonstrated that hemoglobins concentra-
tion changes estimated from PPG could be a valuable additional
tool for the assessment of blood perfusion. The promising com-
parative results with the NIRS system showed that the PPG
boundaries might be effectively extended to NIRS.
REFERENCES
[1] M. Siegemund et al., “Assessment of regional tissue oxygenation,” Inten-
sive Care Med., vol. 25, pp. 1044–1060, 1999.
[2] A. Lima and J. Bakker, “Noninvasive monitoring of peripheral perfusion,”
Intensive Care Med., vol. 31, pp. 1316–1326, 2005.
[3] Y. Sakr, “Techniques to assess tissue oxygenation in the clinical setting,”
Transfusion Apheresis Sci., vol. 43, no. 1, pp. 79–94, 2010.
[4] M. Hickey et al., “A new fibre optic pulse oximeter probe for monitor-
ing splanchnic organ arterial blood oxygen saturation,” Comput. Methods
Programs Biomed., vol. 108, no. 3, pp. 883–888, 2011.
[5] M.E.vanGenderenet al., “Monitoring peripheral perfusion in critically
ill patients at the bedside,” Current Opinion Crit. Care, vol. 8, no. 3,
pp. 273–379, 2012.
[6] P. A. Kyriacou, “Direct pulseoximetry within the esophagus, on the surface
of abdominal viscera, and on free flaps,” Anesthesia Analgesia, vol. 117,
no. 4, pp. 824–833, 2013.
[7] T. Zaman et al., “Pilot investigation of DIEP free flap perfusion utiliz-
ing a multi-wavelength non-invasive optical sensor,” Brit. J. Anaesthesia,
vol. 110, no. 5, pp. 884–885, 2013.
[8] J.T.Moyle,Pulse Oximetry. London, U.K.: BMJ Books, 2002.
[9] P. A. Kyriacou, “Pulse oximetry in the oesophagus,” Physiol. Meas.,
vol. 27, no. 1, pp. R1–R35, 2006.
[10] E. D. Chan et al., “Pulse oximetry: Understanding its basic principles
facilitates appreciation of its limitation,” Respir. Med., vol. 107, no. 6,
pp. 789–799, 2013.
[11] A. A. Kamal et al., “Skin photoplethysmography—A review,” Comput.
Methods Program. Biomed., vol. 28, no. 4, pp. 257–269, 1989.
[12] J. E. Sinex, “Pulse oximetry: Principles and limitations,” Amer. J. Emer-
gency Med., vol. 17, no. 1, pp. 59–66, 1999.
[13] A. Reisner et al., “Utility of photoplethysmography in circulatory moni-
toring,” Anesthesiology, vol. 108, no. 5, pp. 950–958, 2008.
[14] J. Allen, “Photoplethysmography and its application in clinical physio-
logical measurement,” Physiol. Meas., vol. 28, pp. R1–R39, 2007.
[15] A. P. Lima et al., “Use of peripheral perfusion index derived from the
pulse oximetry signal as a noninvasive indicator of perfusion,” Crit. Care
Med., vol. 30, no. 6, pp. 1210–1213, 2002.
[16] P. Zaramella et al., “Foot pulse oximeter perfusion index correlated with
calf muscle perfusion measured by near-infrared spectroscopy in healthy
neonates,” J. Perinatol., vol. 25, no. 6, pp. 417–422, 2005.
[17] K. Shelley, “Photoplethysmography: Beyond the calculation of arterial
oxygen saturation and heart rate,” Anaesthesia Analgesia, vol. 105, no. 6,
pp. S31–S36, 2007.
[18] R. Sahni, “Noninvasive monitoring by photoplethysmography,” Clin. Peri-
natol., vol. 39, no. 3, pp. 573–583, 2012.
[19] M. van Genderen et al., “Peripheral perfusion index as an early predictor
for central hypovolemia in awake healthy volunteers,” Anesthesia Anal-
gesia, vol. 116, no. 2, pp. 351–356, 2013.
[20] V. Kamat, “Pulse oximetry,” Indian J. Anaesthesia, vol. 46, no. 4,
pp. 261–268, 2002.
[21] A. Jubran, “Pulse oximetry,” in Applied Physiology in Intensive Care
Medicine, G. Hedenstierna, J. Mancebo, L. Brochard, and M. R. Pinsky,
Eds. Berlin Germany: Springer-Verlag, 2009, pp. 45–48.
[22] P. A. Kyriacou et al., “Pilot investigation of photoplethysmographic
signals and blood oxygen saturation values during blood pressure cuff-
induced hypoperfusion,” Measurement, vol. 42, pp. 1001–1005, 2009.
[23] M. Shafique et al., “Investigation of photoplethysmographic signals and
blood oxygen saturation values on healthy volunteers during cuff-induced
hypoperfusion using a multimode PPG/SpO2sensor,” Med. Biol. Eng.
Comput., vol. 50, pp. 575–583, 2012.
[24] P. Rolfe, “In vivo near infrared spectroscopy,” Annu. Rev. Biomed. Eng.,
vol. 2, pp. 715–754, 2000.
[25] A. Pellicer and M. del Carmen Bravo, “Near-infrared spectroscopy: A
methodology-focused review,” Seminar Fetal Neonatal Med., vol. 16,
no. 1, pp. 42–49, 2011.
[26] K. McCully and T. Hamaoka, “Near-infrared spectroscopy: What can it
tell us about oxygen saturation in skeletal muscle?” Exercise Sport Sci.
Rev., vol. 28, no. 3, pp. 123–127, 2000.
[27] R. Boushel and C. A. Piantadosi, “Near-infrared spectroscopy for mon-
itoring muscle oxygenation,” Acta Physiologica Scandinavica, vol. 168,
no. 4, pp. 615–622, 2000.
[28] M. Girardis et al., “Muscle perfusion and oxygen consumption by near-
infrared spectroscopy in septic and non-septic-shock patients,” Intensive
Care Med., vol. 29, pp. 1173–1176, 2003.
[29] B. Shadgan et al., “Wireless near-infrared spectroscopy of skeletal muscle
oxygenation and hemodynamics during exercise and ischemia,” Spec-
troscopy, vol. 23, pp. 233–241, 2009.
[30] D. S. Martin et al., “The use of skeletal muscle near infrared spectroscopy
and a vascular occlusion test at high altitude,” High Altitude Med. BIol.,
vol. 14, no. 3, pp. 256–262, 2013.
[31] T. J. Galla et al., “Differentiation between arterial and venous vessel
occlusion by simultaneous measurement with laser Doppler flowmetry
and photoplethysmography,” J. Reconstructive Microsurg., vol. 15, no. 1,
pp. 67–72, 1999.
[32] N. B. Hampson and C. Piantadosi, “Near-infrared monitoring of human
skeletal muscle oxygenation during forearm ischemia,” J. Appl. Physiol.,
vol. 64, no. 6, pp. 2449–2457, 1988.
[33] V. Rybynok et al., “Design and development of a novel multi-channel pho-
toplethysmographic research system,” in IEEE Point-of-Care Healthcare
Technol. Conf., Bangalore, India, 2013, pp. 267–270.
[34] K. Budidha and P. A. Kyriacou, “The human ear canal: Investigation of
its suitability for monitoring photoplethysmographs and arterial oxygen
saturation,” Physiol. Meas., vol. 35, pp. 111–128, 2014.
ABAY AND KYRIACOU: REFLECTANCE PHOTOPLETHYSMOGRAPHY AS NONINVASIVE MONITORING OF TISSUE BLOOD PERFUSION 2195
[35] K. Budidha and P. A. Kyriacou, “Development of an optical probe to
investigate the suitability of measuring photoplethysmographs and blood
oxygen saturation from the human auditory canal,” in Proc. IEEE Annu.
Conf. Eng. Med. Biol. Soc., Osaka, Japan, 2013, pp. 1736–1739.
[36] H. Njoum and P. A. Kyriacou, “Investigation of finger reflectance pho-
toplethysmography in volunteers undergoing a local sympathetic stimu-
lation,” Proc. J. Phys. Conf. Ser., Sensors Appl. XVII, vol. 450, pp. 1–6,
2013.
[37] J. G. Webster, Design of Pulse Oximeters, 1st ed. New York, NY, USA:
Taylor & Francis, 1997.
[38] Y. Mendelson and B. D. Ochs, “Noninvasive pulse oximetry utilizing skin
reflectance photoplethysmography,” IEEE Trans. Biomed. Eng., vol. 35,
no. 10, pp. 798–805, Oct. 1988.
[39] M. Hickey and P. A. Kyriacou, “Optimal spacing between transmitting
and receiving optical fibres in reflectance pulse oximetry,” Proc. J. Phys.
Conf. Ser., vol. 85, no. 1, pp. 1–8, 2007.
[40] Hamamatsu Photonics, “NIRO 200NX measurement principles”,
Hamamatsu, Japan, 2013.
[41] S. Suzuki et al., “A tissue oxygenation monitor using NIR spatially re-
solved spectroscopy,” Proc. SPIE Conf. Opt. Tomography Spectroscopy
Tissue III, San Jose, CA, USA, vol. 3597, 1999, pp. 582–592.
[42] S. Hyttel-Sorensen et al., “Tissue oximetry: A comparison of mean values
of regional tissue saturation, reporducibility and dynamic range of four
NIRS-instruments on the human forearm,” Biomedic Opt. Exp.,vol.2,
no. 11, pp. 3047–3057, 2011.
[43] Y. Mendelson and M. J. McGinn, “Skin reflectance pulse oximetry: In
vivo measurements from the forearm and calf,” J. Clin. Monit.,vol.7,
pp. 7–12, 1991.
[44] P. D. Mannheimer et al., “The influence of large subcutaneous blood
vessels on pulse oximetry,” J. Clin. Monit. Comput., vol. 18, no. 3,
pp. 179–188, 2004.
[45] M. Ferrari et al., “Oxidative metabolism in muscle,” Philosoph. Trans.
Roy. Soc. Biol. Sci., vol. 352, no. 1354, pp. 677–683, 1997.
[46] R. Wariar et al., “A modular NIRS system for clinical measurement
of impaired skeletal muscle oxygenation,” J. Appl. Physiol., vol. 88,
pp. 315–325, 2000.
[47] T. J. Akl et al., “Wireless monitoring of liver hemodynamics in vivo,”
PloS One, vol. 9, no. 7, p. e102396, 2014.
[48] A. Lima and J. Bakker, “Clinical monitoring of peripheral perfusion: There
is more to learn,” Crit. Care, vol. 18, no. 113, pp. 1–3, 2014.
[49] R. A. De Blasi et al., “Comparison of two methods of measuring forearm
oxygen consumption (VO2) by near infrared spectroscopy,” J. Biomed.
Opt., vol. 2, no. 2, pp. 171–175, 1997.
[50] W. F. Ganong, Review of Medical Physiology. East Norwalk, CT, USA:
Appleton & Lange, 1995.
Tomas Ysehak Abay (S’14) was born in Italy in
1986. He received the B.Sc. degree in biomedical en-
gineering from Politecnico di Milano, Milano, Italy,
in 2010. After working for one year as a Medical
Engineer at San Matteo Hospital, Pavia, Italy, he re-
ceived the M.Sc. degree in biomedical engineering
with Healthcare Technology Management from City
University London, London, U.K., in 2012. He is cur-
rently working toward the Ph.D. degree in biomedical
engineering at City University London.
His research interests include photoplethysmog-
raphy, pulse oximetry, near-infrared spectroscopy, and other noninvasive optical
techniques for monitoring tissue blood perfusion.
Panayiotis A. Kyriacou (SM’06) was born in Cyprus
in 1969. He received the B.E.Sc. degree in electrical
engineering from the University of Western Ontario,
London, ON, Canada, and the M.Sc. and Ph.D. de-
grees in medical electronics and physics from St.
Bartholomew’s Medical College, University of Lon-
don, London, U.K.
He is currently a Professor of biomedical engi-
neering and the Associate Dean for Research and
Enterprise at the School of Mathematics Computer
Science and Engineering, City University London,
London. He is also the Director of the Biomedical Engineering Research Cen-
tre. He has authored and coauthored more than 200 publications, including
peer-reviewed journal publications, invited chapters in books, and conference
proceedings. He served as the Chair of the Physiological Measurement Group
and the Chair of the Engineering Advisory Group at the Institute of Physics
and Engineering in Medicine and as the Chair of the Instrument Science and
Technology Group of the Institute of Physics. He is also an Executive Council
Member and Treasurer of the European Alliance for Medical and Biological
Engineering and Science. His main research interests include the understand-
ing, development, and applications of medical instrumentation and sensors to
facilitate the prognosis, diagnosis, and treatment of disease or the rehabilitation
of patients.
