Sensors 23 03723
Sensors 23 03723
Sensors 23 03723
Article
Pulse Oximetry Imaging System Using Spatially Uniform Dual
Wavelength Illumination
Riaz Muhammad 1,† , Kay Thwe Htun 1,† , Ezekiel Edward Nettey-Oppong 1,† , Ahmed Ali 1,2 , Dae Keun Jeon 3 ,
Hyun-Woo Jeong 4 , Kyung Min Byun 5,6, * and Seung Ho Choi 1,7, *
Abstract: Pulse oximetry is a non-invasive method for measuring blood oxygen saturation. However,
its detection scheme heavily relies on single-point measurements. If the oxygen saturation is measured
at a single location, the measurements are influenced by the profile of illumination, spatial variations
in blood flow, and skin pigment. To overcome these issues, imaging systems that measure the
distribution of oxygen saturation have been demonstrated. However, previous imaging systems have
relied on red and near-infrared illuminations with different profiles, resulting in inconsistent ratios
between transmitted red and near-infrared light over space. Such inconsistent ratios can introduce
Citation: Muhammad, R.; Htun, K.T.;
fundamental errors when calculating the spatial distribution of oxygen saturation. In this study, we
Nettey-Oppong, E.E.; Ali, A.; Jeon,
developed a novel illumination system specifically designed for a pulse oximetry imaging system. For
D.K.; Jeong, H.-W.; Byun, K.M.; Choi,
the illumination system, we customized the integrating sphere by coating a mixture of barium sulfate
S.H. Pulse Oximetry Imaging System
and white paint inside it and by coupling eight red and eight near-infrared LEDs. The illumination
Using Spatially Uniform Dual
Wavelength Illumination. Sensors
system created identical patterns of red and near-infrared illuminations that were spatially uniform.
2023, 23, 3723. https://doi.org/ This allowed the ratio between transmitted red and near-infrared light to be consistent over space,
10.3390/s23073723 enabling the calculation of the spatial distribution of oxygen saturation. We believe our developed
pulse oximetry imaging system can be used to obtain spatial information on blood oxygen saturation
Academic Editor: Vittorio Passaro
that provides insight into the oxygenation of the blood contained within the peripheral region of
Received: 3 February 2023 the tissue.
Revised: 25 February 2023
Accepted: 16 March 2023 Keywords: pulse oximetry imaging system; spatially uniform dual wavelength illumination;
Published: 4 April 2023 hyperspectral imaging system
of other important human organs [4,5]. Therefore, accurate and frequent measurements
of oxygen saturation are essential for monitoring the well-being of vital organs and for
the early detection of potential health issues. For measuring blood oxygen saturation, a
variety of options including invasive and non-invasive methods are available from both
the literature and commercial devices. In an invasive SpO2 estimation method, the levels of
oxyhemoglobin (HbO2 ) and deoxyhemoglobin (Hb) can be determined through a blood gas
analysis where blood samples are collected, and the number of oxygen-binding hemoglobin
is counted. Although this technique is accurate, it is costly, time-consuming, and invasive.
Pulse oximetry is commonly used as a non-invasive method for measuring the blood
oxygen saturation (SpO2 ) and the heart rate of patients in hospitals and homes [6]. Pulse
oximetry measures the percentage of oxygenated hemoglobin in the arteries [7]. The
measurements determine the supply of oxygen to peripheral tissues and the efficiency
of oxygenation in the pulmonary alveoli [8,9]. During detection, the interaction of light
with the area of measurement is achieved via transmission or reflection methods. In the
transmission method, light incidents on one side of the targeted peripheral region, and the
transmitted light is detected by a photodetector placed either on the opposite side or side-
by-side with the light source, whereas the light source and photodetector are located on
the same side of the targeted region in the reflection method [10,11]. Photoplethysmogram
(PPG) signals, generated by shining red and near-infrared light on the skin, provide
information on both oxygenated and deoxygenated hemoglobin and are commonly used
to quantify blood oxygen saturation. This is achieved by evaluating the ratio of absorption
coefficients of oxyhemoglobin to deoxyhemoglobin, which reflects the saturation level of
oxygen in the blood [12].
Conventional pulse oximetry methods are limited to single-point measurements, and
hence the measurements are influenced by the shape of illumination, spatial variations
in blood flow [13], and skin pigment [14,15]. To overcome these challenges, imaging
systems that measure the distribution of blood oxygen saturation have been demonstrated.
Humphreys and Markham [16] presented a contactless system equipped with a CMOS
camera to capture photoplethysmography (PPG) signals at 760 nm and 880 nm, respectively.
Two lights were alternately illuminated on the fingertip to detect rhythmic arterial pulsation.
Subsequently, the results were compared to those of a conventional contact-based pulse
oximeter. Although the results were comparable, the SNR of the PPG signal was highly
dependent on the location in the peripheral region of the tissue. Al-Naji et. al. [17] have
used the green and red channels of a digital camera to estimate SpO2 values. However, the
estimated SpO2 values are greatly influenced by the lighting conditions.
Moreover, previous camera-based pulse oximetry imaging systems relied on two
different patterns of red and near-infrared illuminations [18–20]. When spatial profiles of
the input red and near-infrared illumination are different, the ratio between transmitted
red and near-infrared light also differs across space [20]. The mismatch between the red
and near-infrared illumination profiles can introduce fundamental errors [21]. Therefore,
spatially uniform dual wavelength illumination is required to eliminate the occurrence of
artifacts due to spatially varying light sources.
In this study, we developed a novel illumination system specifically designed for use
in a pulse oximetry imaging system. The illumination system developed in this study
created two identical patterns of red (630 nm) and near-infrared (940 nm) illuminations that
are spatially uniform. To achieve this unique illumination, we customized an integrating
sphere by coating its interior with a mixture of barium sulfate and white paint.
An integrating sphere, also referred to as the Ulbricht sphere, is an optical device used
for the spatial integration of radiant flux, and thus has a diffusing effect or the uniform
scattering of light. It is composed of a hollow sphere with small openings which serve as
input and output ports for light. The interior of the sphere is coated with a highly reflective
material to obtain diffusely reflecting walls, which enables multiple scattering reflections of
light within the sphere. Incident light rays are scattered multiple times within the sphere,
effectively averaging out any directional variations in the light output. Consequently,
Sensors 2023, 23, 3723 3 of 18
Z2π π/2
Z
p(θ, φ) = f (θ,φ0 ; θ 0 , φ0 )sin θ 0 cos θ 0 dθ 0 dφ0 (2)
0 0
For an internally illuminated integrating sphere, both the numerous surface reflections
and losses via the port apertures required for the input flux are considered in the radiance
equation [22].
Φi p
L= × (3)
π As 1 − p (1 − f )
where f is the port fraction. The radiance equation has two components; the first component
is identical to the general expression for radiance (Equation (1)), and the second component
is termed the sphere multiplier [22]. The sphere multiplier, M, is a unitless quantity that
accounts for the radiance increase resulting from multiple reflections.
p
M= (4)
1 − p (1 − f )
The value of M is determined by both the reflectance of the sphere surface and the
port fraction. The port fraction is expressed as a function of the input port area, Ai , the
output port area, Ao , and the sphere wall area, As .
Ai + A o
f = (5)
As
The developed pulse oximetry imaging system incorporates this illumination system
to capture high-precision spatial and temporal data. Using a hyperspectral imaging ap-
proach, we obtained spectral information from different locations within the peripheral
region based on the acquired spatio-temporal data. Hyperspectral imaging is a multivariate
imaging technique. An image made of I rows and J columns acquired over K variables is a
classical multivariate image. The commonly utilized variables are wavelengths, but other
variables can be used as well [24]. By interpolating between the coinciding points of the
dual images captured due to the dual wavelength light source, a spectrum is generated to
characterize different locations within the peripheral region. Utilizing the hyperspectral
imaging approach and fundamental theory of pulse oximetry, a spatial map of the blood
oxygen saturation was produced. Moreover, the developed system overcomes the limita-
tions of existing systems, where the spatial intensity distribution of red and near-infrared
lights is not uniform, yielding approximate blood oxygen saturation measurements.
Sensors 2023, 23, 3723 4 of 18
2.1.5. Microcontroller for Generating Trigger Signals for Camera and Light Source
An Arduino microcontroller was used to control both the red (630 nm) and near-
infrared (940 nm) LEDs and the camera, as shown in Figure 1a. A separate power supply
was required for the 630 nm LEDs due to the higher current draw, which is shown in
Figure 1b. This power supply provided a regulated 12 volts and was switched on and
off by a trigger signal generated by the Arduino through a transistor (model MPS 2222A,
ON Semiconductor, Scottsdale, AZ, USA). In contrast, the 940 nm LED array was pow-
ered directly from the trigger signal generated by the Arduino, eliminating the need for
an additional power supply. Using the Arduino microcontroller to generate trigger sig-
nals ensures precise control of both red (630 nm) and near-infrared (940 nm) LEDs for
accurate imaging.
[HbO2 ]
SpO2 = ×100% (6)
[HbO2 ]+[Hb]
where [HbO2 ] and [Hb] are the concentrations of hemoglobin with oxygen and without
oxygen, respectively.
The specific absorption coefficients of oxygenated (HbO2) and deoxygenated hemoglobin
(Hb) have distinct spectral dependence [32], as included in Figure 1e. The difference
between these coefficients allows for the quantification of each constituent, forming the
basis of pulse oximetry [33,34]. At wavelengths below 800 nm, the Hb-specific absorp-
tion coefficients are higher than the HbO2 one, while in the region above 800 nm, the
HbO2 absorptions dominate. As shown in Figure 1e, red light at 630 nm is absorbed
by deoxygenated hemoglobin (Hb) more than oxygenated hemoglobin (HbO2 ), whereas
near-infrared light at 940 nm is absorbed by HbO2 more than Hb. Upon transmission or
reflection through tissue, the red or near-infrared light is attenuated due to absorption by
Sensors 2023, 23, 3723 7 of 18
biological molecules and scattering by larger tissue structures. The modified Beer-Lambert
Law [35] is used to express the intensity of the transmitted or reflected light from the skin
tissue to quantify the light absorption by hemoglobin as follows:
It = Io e−CDα (7)
where It is the intensity of the transmitted light, Io is the intensity of incident light, C is
the concentration of the sample, α is the absorption coefficient of the tissue, and D is the
optical path length (i.e., the distance traveled through the sample). The calculation of
SpO2 based on pulse oximetry assumes that the pulsatile component of optical absorp-
tion is due to the pulsating flow of arterial blood, while the non-pulsatile component
stems from non-pulsating arterial blood, venous blood, and other tissues. The ratio of ab-
sorbances is calculated using the red (630 nm) and near-infrared (940 nm) time signals in the
following equation:
AC630
R630 =
DC630
AC940
R940 = (8)
DC940
R630
RR =
R940
where AC is the pulsatile component, DC is the non-pulsatile component, and RR is
the ratio of ratios of the absorbances at the two wavelengths. Previous studies [36,37]
have shown that the pulsatile and non-pulsatile components correspond to the standard
deviation and mean color intensities of the red and near-infrared frames, respectively. A
nearly linear relationship exists between SpO2 values and RR [38]. Therefore, the value of
SpO2 is estimated from RR according to the equation:
where m and c are empirically determined through linear regression for each volunteer,
m is the slope of the estimated regression line, and c is the y-intercept [39]. The empirical
approximation technique is used to correct errors in the measured values caused by the
assumption of only two substances in the light path [40].
Figure 2. Image acquisition and data processing for SpO2 calculation: Image acquisition. The image
data is captured alternately at 630 nm and 940 nm in a time series of frames, with a resolution of
1280 × 1024 pixels for each frame. The region above the output port of the integrating sphere was
considered a valid measurement area. Image data processing. The panel illustrates the steps for
calculating the SpO2 values and displays the AC and DC components of the 630 nm and 940 nm
photoplethysmogram (PPG) signals.
The PPG signals were used as the physiological indicators to analyze the pulsatile
changes in blood volume and blood oxygen saturation levels at various locations over
different wavelengths. Each of the two PPG signals was divided into a 10-s subset, and
the AC and DC components of the PPG signals were obtained using average peak-to-peak
and mean values, respectively, from each of these subsets (see Figure 2). The minimum
points at each signal were interpolated across all spatial locations to remove the baseline
from the PPG signals, and the DC component was subtracted from the AC component.
Consequently, the RR values were calculated from Equation (8) using the determined AC
and DC components of the PPG signals. Finally, the blood oxygen saturation (SpO2 ) values
were extracted from the calculated RR values at each spatial location using Equation (9).
Figure
Figure 4.4. Quantification
Quantificationof of
uniformity
uniformity and similarity
and of red
similarity andand
of red near-infrared illuminations:
near-infrared illuminations:(a)
Transmitted
(a) Transmittedlightlight
illumination intensity
illumination at 630atnm
intensity 630(left), 940 nm
nm (left), (center),
940 and theand
nm (center), intensity differ-
the intensity
ence between 630 nm and 940 nm (right) emitted from the output port of the integrating sphere. To
difference between 630 nm and 940 nm (right) emitted from the output port of the integrating sphere.
capture the images of the transmitted light illumination intensity, a diffuser was placed over the
To capture the images of the transmitted light illumination intensity, a diffuser was placed over the
output port of the sphere with a field of view (FOV) of 1.8 cm on the x-axis and 1.4 cm on the y-axis;
output
(b) port of thedisplays
The spectrum sphere with a field in
the change of intensity
view (FOV) of 1.8 cm
difference between x-axis
on the 630 nmand
and1.4
940cmnm onatthe y-axis;
a central
(b) The spectrum displays the change in intensity difference between 630 nm and 940
field of view of 1.2 cm over a period of time ranging from 72 ms to 720 ms; (c) The plot depicts thenm at a central
field of view ofin
mean-variance 1.2 cm overdifference
intensity a period of time ranging
between 630 nmfrom
and 72940ms
nmtoover
720 ms;
time.(c) The plot depicts the
mean-variance in intensity difference between 630 nm and 940 nm over time.
3.2. Photoplethysmography
3.2. Photoplethysmography
An experimental test was conducted by imaging a fingertip (refer to supplementary
An experimental test was conducted by imaging a fingertip (refer to supplementary
Video S1). Figure 5a illustrates the relative position of the imaged illuminated area of the
Video S1). Figure 5a illustrates the relative position of the imaged illuminated area of
the fingertip. The mean values of the pixels contained in the highlighted regions of
the fingertip are plotted against time. The waveforms in the projected PPG graph are
inverted, and the received pixel light intensity is replaced with light absorption on the
vertical axis. Light absorption is directly proportional to the peripheral arterial pressure
waveform, whereas the received light intensity is inversely proportional. In the analyzed
PPG signal, the systolic peaks and diastolic troughs are easily distinguishable. Moreover,
the dicrotic notch—an inflection in the waveform caused by the abrupt closure of the
aortic valve—is distinctly visible. The outlined observations are consistent with expected
physiological behavior and further support the validity of the PPG signal as a measure of
cardiovascular activity.
Light absorption is directly proportional to the peripheral arterial pressure waveform,
whereas the received light intensity is inversely proportional. In the analyzed PPG signal,
the systolic peaks and diastolic troughs are easily distinguishable. Moreover, the dicrotic
notch—an inflection in the waveform caused by the abrupt closure of the aortic valve—is
distinctly visible. The outlined observations are consistent with expected physiological
Sensors 2023, 23, 3723 12 of 18
behavior and further support the validity of the PPG signal as a measure of cardiovascular
activity.
Figure 5.
Figure 5. Measurement
Measurementofofphotoplethysmogram
photoplethysmogram using pulse
using oximetry
pulse imaging
oximetry system:
imaging (a) The
system: (a)pho-
The
toplethysmogram (PPG) signals acquired at various spatial locations are displayed, demonstrating
photoplethysmogram (PPG) signals acquired at various spatial locations are displayed, demonstrating
time-varying spatial features at different wavelengths; (b) the signal-to-noise ratio (SNR) plot for
time-varying spatial features at different wavelengths; (b) the signal-to-noise ratio (SNR) plot for
both the red and near-infrared wavelengths; (c) Fast Fourier Transform (FFT) spectra of the RED
both the red
(630 nm) andand near-infrared
near-infrared wavelengths;
(NIR) (940 nm) PPG(c) Fast Fourier
signals Transform
are shown, (FFT) spectra
highlighting of therate
the heart REDat
(630 nm) and near-infrared (NIR) (940 nm) PPG signals are
1.19 Hz (71.2 bpm) and respiratory rate at 0.25 Hz (15 BR/min).shown, highlighting the heart rate at
1.19 Hz (71.2 bpm) and respiratory rate at 0.25 Hz (15 BR/min).
The signal-to-noise ratio (SNR) plot suggests that the PPG signals obtained from dif-
The signal-to-noise ratio (SNR) plot suggests that the PPG signals obtained from
ferent spatial locations are consistent and have a relatively constant SNR (see Figure 5b).
different spatial locations are consistent and have a relatively constant SNR (see
This consistency across spatial locations, distinct pulse amplitude, and phase information
Figure 5b). This consistency across spatial locations, distinct pulse amplitude, and
demonstrates the efficiency of the pulse oximetry imaging system and further confirms
phase information demonstrates the efficiency of the pulse oximetry imaging system
the uniformity of the illumination used. A high-quality PPG signal which reflects the
and further confirms the uniformity of the illumination used. A high-quality PPG
amount of transmitted or reflected light that penetrates the skin is crucial for obtaining
signal which reflects the amount of transmitted or reflected light that penetrates the
accurate physiological parameters such as heart rate, respiratory rate, and blood oxygen
skin is crucial for obtaining accurate physiological parameters such as heart rate, respi-
saturation (SpO2). Fourier spectra of the red (630 nm) and NIR (940 nm) PPG signals are
ratory rate, and blood oxygen saturation (SpO2 ). Fourier spectra of the red (630 nm)
depicted in Figure 5c. The pulsatile component, evident at 1.19 Hz (71.4 bpm), is profound
and NIR (940 nm) PPG signals are depicted in Figure 5c. The pulsatile component,
in both spectra. The respiratory component, demonstrated at 0.25 Hz (15 BR/min), is also
evident at 1.19 Hz (71.4 bpm), is profound in both spectra. The respiratory compo-
apparent
nent, in both signals.
demonstrated The acquired
at 0.25 results areis
Hz (15 BR/min), comparable to thein
also apparent results
both obtained
signals. from
The
acquired results are comparable to the results obtained from a Patient monitor (model
M40, MEDIANA Co., Ltd., Seoul, South Korea), which recorded 1.23 Hz (73 bpm) and
0.26 Hz (16 BR/min) for the pulsatile and respiratory components, respectively.
Table 1 provides a comparative analysis between the proposed system and other pulse
oximetry imaging systems. Previously demonstrated pulse oximetry imaging systems
have used LED arrays [39,43] or LED rings [16,44] that relied on two different illumination
profiles for dual wavelengths. When the spatial profiles of the two illuminations are
different, the ratio between transmitted light will also differ. Such a mismatch between
the illumination profiles can introduce fundamental errors when calculating the spatial
distribution of blood oxygen saturation. In contrast, our illumination system created two
Sensors 2023, 23, 3723 13 of 18
identical patterns of red and near-infrared illumination, of which patterns were spatially
uniform with standard error of only 1.2% and 1.3%, respectively (see Figure 3b). We utilized
this illumination system to capture spatial and temporal data, from which we obtained
spectral information at different locations within the peripheral region.
Table 1. Comparison of our illumination system with other reported illumination systems.
Figure 6. Spatio-temporal assessment of blood oxygen saturation using pulse oximetry imaging
system: (a) The image illustrates the validated area of the finger illuminated by red (630 nm). The
raw Photoplethysmogram (PPG) waveform is shown for three cardiac cycles. The figure also shows
the processed 12 images for a single cardiac cycle showing spatial variation in pulsatile blood flow;
Sensors 2023, 23, 3723 15 of 18
(b) The image illustrates the validated area of the finger illuminated by red (940 nm). The raw
Photoplethysmogram (PPG) waveform is shown for three cardiac cycles. The figure also shows the
processed 12 images for a single cardiac cycle showing spatial variation in pulsatile blood flow (for
a video demonstration of three cycles, refer to Supplementary Materials Videos S2 and S3); (c) The
SpO2 map represents the spatial distribution of blood oxygen saturation within the validated region
of measurement. The average SpO2 measured for the subject was 97.17%.
SpO2 Quantification
To evaluate the performance of our illumination system, we calculated the spatial
distribution of blood oxygen saturation (SpO2 ) in the fingertip using the fundamental theory
of pulse oximetry. In pulse oximetry, the AC component corresponds to the maximum
peak of the PPG signal. The time-varying spatial maps of pulsatile blood flow obtained
using both red (630 nm) and near-infrared (940 nm) images correspond to the amplitude
variations of a PPG signal (see Figure 6a,b). The processed images under red illumination
show the maximum peak (AC630 ) of the PPG signal, which corresponds to high blood
volume at t = 684 ms. Similarly, under near-infrared illumination, the AC940 corresponds to
high blood volume at t = 720 ms. We normalized the AC components with DC components
at these temporal points. Last, we calculated SpO2 distribution using Equation (4), and by
utilizing calibrated m and c values of −21.56 and 110.66, respectively. The average SpO2
level measured by our system was 97.15% (see Figure 6c). For comparison, the SpO2 level
of the subject was also measured using the Patient monitor (model M40, MEDIANA Co.,
Ltd., Wonju, Republic of Korea), and a value of 99% was recorded. To minimize potential
calibration errors, it is recommended that the system be tested on several volunteers. The
developed pulse oximetry imaging system successfully measured the pulsatile variation
in blood and SpO2 levels of the peripheral region. Table 2 compares the SpO2 , heart rate,
and respiratory rate data from our proposed system and a commercially available Patient
monitor device (model M40, MEDIANA Co., Ltd., Wonju, Republic of Korea).
Table 2. Comparison of physiological parameters from our system with a commercially available
patient monitoring system.
Table 3 compares the measurement systems for blood oxygen saturation based on
the parameters analyzed. Single channel pulse oximeters only monitor temporal changes
in transmitted light at a single point to calculate the blood oxygen saturation. If blood
oxygen saturation is measured at a single location, the measurements are influenced by
the shape of illumination and spatial variations in blood flow [2]. However, our proposed
imaging system can capture the spatial and temporal changes in transmitted light at
various spatial locations over the red (I630 nm see Figure 6a) and near-infrared (I940 nm see
Figure 6b). Additionally, the proposed system can provide the spatial distribution and
temporal change of SpO2 , as demonstrated by Figure 6c. On the other hand, the single-
channel pulse oximeter provides only the temporal changes of I630 nm and I940 nm , and the
SpO2 at a single location. The table highlights the superior performance of our proposed
system over single-channel pulse oximeters in terms of spatial and spectral data acquisition.
Sensors 2023, 23, 3723 16 of 18
Table 3. Comparison of our system’s specifications with a single-channel pulse oximetry system.
System Parameter Our Proposed System Single Channel Pulse Oximetry System
Spatial resolution 106 × 106 µm N/A
Provide spatial distribution and temporal
change of I630 nm ;
Spectral data at Provide temporal change of I630 nm at a
Series of I630 nm images visualizes
630 nm, I630 nm single location
pulsatile blood flow on region of interest
(Figure 6a).
Provide spatial distribution and temporal
change of I940 nm ;
Spectral data at Provide temporal change of I940 nm at a
Series of I940 nm images visualizes
940 nm, I940 nm single location
pulsatile blood flow on region of interest
(Figure 6b).
Provide spatial distribution and temporal Provide temporal change of SpO2 at a
Blood oxygen saturation level, SpO2
change of SpO2 (Figure 6c) single location
4. Conclusions
In this study, we developed a camera-based pulse oximetry imaging system that
employs a spatially uniform dual wavelength light source to observe the spatial-temporal
changes in blood oxygen saturation over the peripheral regions, such as the fingertip or
earlobe. We used a pulse oximetry technique in combination with a hyper-spectral imaging
approach to characterize different locations within the peripheral region. Our results
showed that both 630 nm and 940 nm images contain physiological information about the
blood flow within the peripheral region of the tissue under investigation. We combined
the processed red and near-infrared images to estimate the blood oxygen saturation at
individual locations within the peripheral region. The measured physiological parameters
were comparable to those of a commercial Patient monitor.
Our pulse oximetry imaging system offers several advantages over conventional single-
channel pulse oximetry methods. Our imaging technique eliminates limitations related to
inconsistent illumination profiles between the two wavelengths in the peripheral region
and provides rich spatio-temporal information about blood oxygen saturation at different
locations. The development of an integrated device utilizing this imaging technique holds
significant potential for improving biomedical applications, such as wound monitoring and
real-time, long-term monitoring of blood oxygen saturation, where accurate tissue oxygen
content information is critical.
Supplementary Materials: The following supporting information can be downloaded at: https:
//www.mdpi.com/article/10.3390/s23073723/s1, Video S1: Real time recorded video by conducting
the experiment, Video S2: The visualization of pulsatile blood flow by using 940 nm, Video S3: The
visualization of pulsatile blood flow by using 630 nm.
Author Contributions: Conceptualization, S.H.C., R.M., A.A. and D.K.J.; Methodology, S.H.C., R.M.,
K.T.H., E.E.N.-O. and K.M.B.; Conducting experiments, R.M., K.T.H., E.E.N.-O. and A.A.; Writing-
original draft preparation, S.H.C., R.M., K.T.H., E.E.N.-O. and H.-W.J.; Writing—review & editing,
S.H.C., R.M., K.T.H., E.E.N.-O. and D.K.J.; Formal analysis, S.H.C., R.M., D.K.J., A.A., H.-W.J. and
K.M.B.; Project administration, S.H.C., A.A., D.K.J., H.-W.J. and K.M.B.; Funding acquisition, S.H.C.
and K.M.B. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by the National Research Foundation of Korea (NRF),
grant-funded by the Korean government (MSIT) (No. 2022R1A2C1010151; 2022R1C1C1011328;
2022H1D3A2A02081592) and the Brain Korea 21 Four Program. This research was also supported
by “Regional Innovation Strategy (RIS)” through the National Research Foundation of Korea
(NRF) funded by the Ministry of Education (MOE) in 2023 (2022RIS-005).
Institutional Review Board Statement: Ethical review and approval were waived for this study, due
to non-invasive measurement.
Sensors 2023, 23, 3723 17 of 18
Informed Consent Statement: Informed consent was obtained from all subjects involved in
the study.
Data Availability Statement: The data presented in this study are available from the corresponding
authors upon request.
Conflicts of Interest: The authors declare no conflict of interest.
References
1. Manta, C.; Jain, S.S.; Coravos, A.; Mendelsohn, D.; Izmailova, E.S. An evaluation of biometric monitoring technologies for vital
signs in the era of COVID-19. Clin. Transl. Sci. 2020, 13, 1034–1044. [CrossRef] [PubMed]
2. Tamura, T. Current progress of photoplethysmography and SPO2 for health monitoring. Biomed. Eng. Lett. 2019, 9, 21–36.
[CrossRef] [PubMed]
3. Strapazzon, G.; Gatterer, H.; Falla, M.; Cappello, T.D.; Malacrida, S.; Turner, R.; Schenk, K.; Paal, P.; Falk, M.; Schweizer, J.; et al. Hypoxia
and hypercapnia effects on cerebral oxygen saturation in avalanche burial: A pilot human experimental study. Resuscitation 2021,
158, 175–182. [CrossRef] [PubMed]
4. Mas-Bargues, C.; Sanz-Ros, J.; Román-Domínguez, A.; Inglés, M.; Gimeno-Mallench, L.; El Alami, M.; Viña-Almunia, J.; Gambini,
J.; Viña, J.; Borrás, C. Relevance of oxygen concentration in stem cell culture for regenerative medicine. Int. J. Mol. Sci. 2019,
20, 1195. [CrossRef]
5. Yin, Y.; Shu, S.; Qin, L.; Shan, Y.; Gao, J.-H.; Lu, J. Effects of mild hypoxia on oxygen extraction fraction responses to brain
stimulation. J. Cereb. Blood Flow Metab. 2021, 41, 2216–2228. [CrossRef]
6. Michard, F. Hemodynamic monitoring: Would a pulse oximeter do the job? Crit. Care Med. 2021, 49, 383–386. [CrossRef]
7. Jubran, A. Pulse oximetry. Intensive Care Med. 2004, 30, 2017–2020. [CrossRef]
8. Ferrando, C.; Tusman, G.; Suarez-Sipmann, F.; León, I.; Pozo, N.; Carbonell, J.; Puig, J.; Pastor, E.; Gracia, E.;
Gutiérrez, A.; et al. Individualized lung recruitment maneuver guided by pulse-oximetry in anesthetized patients un-
dergoing laparoscopy: A feasibility study. Acta Anaesthesiol. Scand. 2018, 62, 608–619. [CrossRef]
9. Amouzeshi, A.; Salehi, F.; Ehsani, H.; Jomefourjan, S.; Jani, M.; Amouzeshi, Z. Comparison of pulse oximetry and clinical
examination for proper tissue perfusion in congenital heart patients. J. Surg. Trauma 2021, 9, 26–31.
10. Matsushita, K.; Aoki, K.; Kakuta, N.; Yamada, Y. Fundamental study of reflection pulse oximetry. Opt. Rev. 2003, 10, 482–487.
[CrossRef]
11. Grap, M.J. Pulse oximetry. Crit. Care Nurse 2002, 22, 69–74. [CrossRef]
12. Pothisarn, W.; Chewpraditkul, W.; Yupapin, P. Noninvasive hemoglobin-measurement-based pulse oximetry. In Optics in Health
Care and Biomedical Optics: Diagnostics and Treatment; SPIE: Bellingham, WA, USA, 2002.
13. Kumar, M.; Suliburk, J.W.; Veeraraghavan, A.; Sabharwal, A. PulseCam: A camera-based, motion-robust and highly sensitive
blood perfusion imaging modality. Sci. Rep. 2020, 10, 4825. [CrossRef]
14. Cabanas, A.M.; Fuentes-Guajardo, M.; Latorre, K.; León, D.; Martín-Escudero, P. Skin pigmentation influence on pulse oximetry
accuracy: A systematic review and bibliometric analysis. Sensors 2022, 22, 3402. [CrossRef]
15. Keller, M.D.; Harrison-Smith, B.; Patil, C.; Arefin, M.S. Skin Colour Affects the Accuracy of Medical Oxygen Sensors; Nature Publishing
Group: UK, London, 2022.
16. Humphreys, K.; Markham, C.; Ward, T.E. A CMOS camera-based system for clinical photoplethysmographic applications. In
Opto-Ireland 2005: Imaging and Vision; SPIE: Bellingham, WA, USA, 2005.
17. Al-Naji, A.; Khalid, G.; Mahdi, J.; Chahl, J. Non-contact SpO2 prediction system based on a digital camera. Appl. Sci. 2021,
11, 4255. [CrossRef]
18. Moco, A.V.; Stuijk, S.; De Haan, G. Ballistocardiographic artifacts in PPG imaging. IEEE Trans. Biomed. Eng. 2015, 63, 1804–1811.
[CrossRef]
19. Moço, A.V.; Stuijk, S.; de Haan, G. Motion robust PPG-imaging through color channel mapping. Biomed. Opt. Express 2016, 7,
1737–1754. [CrossRef]
20. Mannheimer, P.D. The light–tissue interaction of pulse oximetry. Anesth. Analg. 2007, 105, S10–S17. [CrossRef]
21. Torp, K.D.; Modi, P.; Simon, L.V. Pulse oximetry. In StatPearls; StatPearls Publishing: St. Petersburg, FL, USA, 2022.
22. Labsphere. Technical Guide: Integrating Sphere Theory and Applications; Labsphere North Sutton: Sutton, NH, USA, 2013.
23. Edwards, D.K.; Gier, J.T.; Nelson, K.E.; Roddick, R.D. Integrating sphere for imperfectly diffuse samples. JOSA 1961, 51, 1279–1288.
[CrossRef]
24. Geladi, P.; Burger, J.; Lestander, T. Hyperspectral imaging: Calibration problems and solutions. Chemom. Intell. Lab. Syst. 2004, 72,
209–217. [CrossRef]
25. Sun, Y.; Thakor, N. Photoplethysmography revisited: From contact to noncontact, from point to imaging. IEEE Trans. Biomed. Eng.
2015, 63, 463–477. [CrossRef]
26. Taylor-Williams, M.; Spicer, G.; Bale, G.; Bohndiek, S.E. Noninvasive hemoglobin sensing and imaging: Optical tools for disease
diagnosis. J. Biomed. Opt. 2022, 27, 080901. [CrossRef] [PubMed]
27. Torrent, J.; Barrón, V. Diffuse reflectance spectroscopy. In Methods of Soil Analysis Part 5—Mineralogical Methods; SSSA: Madison,
WI, USA, 2008; Volume 5, pp. 367–385.
Sensors 2023, 23, 3723 18 of 18
28. Vaqar, A.; Haque IR, I.; Zaidi, T. Spectroscopic Properties of Blood for Pulse Oximeter Design. In Proceedings of the 2019 9th
International Conference on Biomedical Engineering and Technology, Tokyo, Japan, 28–30 March 2019.
29. Caramizoiu, S.; Mihalache, I. Highly reflective surface coating using BaSO4 . In Proceedings of the 2022 International Semiconduc-
tor Conference (CAS), Poiana Brasov, Romania, 12–14 October 2022; IEEE: Piscataway, NJ, USA, 2022.
30. Dias, L.D.S.; Junior, J.C.D.S.; Felicio, A.L.D.S.M.; de Franca, J.A. A NIR photometer prototype with integrating sphere for the
detection of added water in raw milk. IEEE Trans. Instrum. Meas. 2018, 67, 2812–2819. [CrossRef]
31. Nitzan, M.; Romem, A.; Koppel, R. Pulse oximetry: Fundamentals and technology update. Med. Devices Evid. Res. 2014, 7,
231–239. [CrossRef] [PubMed]
32. Smuda, K.; Gienger, J.; Hönicke, P.; Neukammer, J. Function of hemoglobin-based oxygen carriers: Determination of methe-
moglobin content by spectral extinction measurements. Int. J. Mol. Sci. 2021, 22, 1753. [CrossRef] [PubMed]
33. Casalino, G.; Castellano, G.; Zaza, G. A mHealth solution for contact-less self-monitoring of blood oxygen saturation. In
Proceedings of the 2020 IEEE Symposium on Computers and Communications (ISCC), Rennes, France, 7–10 July 2020; IEEE:
Piscataway, NJ, USA, 2020.
34. Rosa, A.d.F.G.; Betini, R.C. Noncontact SpO2 measurement using Eulerian video magnification. IEEE Trans. Instrum. Meas. 2019,
69, 2120–2130. [CrossRef]
35. Kocsis, L.; Herman, P.; Eke, A. The modified Beer–Lambert law revisited. Phys. Med. Biol. 2006, 51, N91. [CrossRef]
36. Selvaraju, V.; Spicher, N.; Wang, J.; Ganapathy, N.; Warnecke, J.M.; Leonhardt, S.; Swaminathan, R.; Deserno, T.M. Continuous
monitoring of vital signs using cameras: A systematic review. Sensors 2022, 22, 4097. [CrossRef]
37. Labati, R.D.; Piuri, V.; Rundo, F.; Scotti, F.; Spampinato, C. Biometric recognition of PPG cardiac signals using transformed
spectrogram images. In Pattern Recognition, Proceedings of the ICPR International Workshops and Challenges, Virtual Event, 10–15
January 2021; Springer: Berlin/Heidelberg, Germany, 2021.
38. Tian, X.; Wong, C.-W.; Ranadive, S.M.; Wu, M. A Multi-Channel Ratio-of-Ratios Method for Noncontact Hand Video Based SpO $
_2 $ Monitoring Using Smartphone Cameras. IEEE J. Sel. Top. Signal Process. 2022, 16, 197–207. [CrossRef]
39. Shao, D.; Liu, C.; Tsow, F.; Yang, Y.; Du, Z.; Iriya, R.; Yu, H.; Tao, N. Noncontact monitoring of blood oxygen saturation using
camera and dual-wavelength imaging system. IEEE Trans. Biomed. Eng. 2015, 63, 1091–1098. [CrossRef]
40. Wu, J.; Gunther, J.; Jayet, B.; Ray, N.; Andersson-Engels, S.; Kainerstorfer, J.M. Self-calibrated pulse oximetry based on absorption
changes in photon pathlength. In Proceedings of the European Conference on Biomedical Optics, Munich, Germany, 20–24 June
2021; Optica Publishing Group: Washington, DC, USA, 2021.
41. Yuan, S.; Devor, A.; Boas, D.A.; Dunn, A.K. Determination of optimal exposure time for imaging of blood flow changes with laser
speckle contrast imaging. Appl. Opt. 2005, 44, 1823–1830. [CrossRef]
42. Fercher, A.; Briers, J. Flow visualization by means of single-exposure speckle photography. Opt. Commun. 1981, 37, 326–330.
[CrossRef]
43. Sun, Y.; Hu, S.; Azorin-Peris, V.; Kalawsky, R.; Greenwald, S. Noncontact imaging photoplethysmography to effectively access
pulse rate variability. J. Biomed. Opt. 2013, 18, 061205. [CrossRef]
44. Fan, Q.; Li, K. Noncontact imaging plethysmography for accurate estimation of physiological parameters. J. Med. Biol. Eng. 2017,
37, 675–685. [CrossRef]
45. Okubo, K. NIR hyperspectral imaging. In Transparency in Biology: Making the Invisible Visible; Springer: Berlin/Heidelberg,
Germany, 2021; pp. 203–222.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.