ArticlePDF AvailableLiterature Review

Review of short-wave infrared spectroscopy and imaging methods for biological tissue characterization

Authors:

Abstract and Figures

We present a review of short-wave infrared (SWIR, defined here as similar to 1000 to 2000 nm) spectroscopy and imaging techniques for biological tissue optical property characterization. Studies indicate notable SWIR absorption features of tissue constituents including water (near 1150, 1450, and 1900 nm), lipids (near 1040, 1200, 1400, and 1700 nm), and collagen (near 1200 and 1500 nm) that are much more prominent than corresponding features observed in the visible and near-infrared (VIS-NIR, defined here as similar to 400 to 1000 nm). Furthermore, the wavelength dependence of the scattering coefficient has been observed to follow a power-law decay from the VIS-NIR to the SWIR region. Thus, the magnitude of tissue scattering is lower at SWIR wavelengths than that observed at VIS or NIR wavelengths, potentially enabling increased penetration depth of incident light at SWIR wavelengths that are not highly absorbed by the aforementioned chromophores. These aspects of SWIR suggest that the tissue spectroscopy and imaging in this range of wavelengths have the potential to provide enhanced sensitivity (relative to VIS-NIR measurements) to chromophores such as water and lipids, thereby helping to characterize changes in the concentrations of these chromophores due to conditions such as atherosclerotic plaque, breast cancer, and burns. (C) 2015 Society of Photo-Optical Instrumentation Engineers (SPIE)
Content may be subject to copyright.
Review of short-wave infrared
spectroscopy and imaging methods
for biological tissue characterization
Robert H. Wilson
Kyle P. Nadeau
Frank B. Jaworski
Bruce J. Tromberg
Anthony J. Durkin
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/02/2015 Terms of Use: http://spiedl.org/terms
Review of short-wave infrared spectroscopy and
imaging methods for biological tissue
characterization
Robert H. Wilson,aKyle P. Nadeau,aFrank B. Jaworski,bBruce J. Tromberg,aand Anthony J. Durkina,*
aUniversity of California, Irvine, Beckman Laser Institute, 1002 Health Sciences Road, Irvine, California 92612, United States
bRaytheon Vision Systems, 75 Coromar Drive, Goleta, California 93117, United Sates
Abstract. We present a review of short-wave infrared (SWIR, defined here as 1000 to 2000 nm) spectroscopy
and imaging techniques for biological tissue optical property characterization. Studies indicate notable SWIR
absorption features of tissue constituents including water (near 1150, 1450, and 1900 nm), lipids (near
1040, 1200, 1400, and 1700 nm), and collagen (near 1200 and 1500 nm) that are much more prominent
than corresponding features observed in the visible and near-infrared (VIS-NIR, defined here as 400 to
1000 nm). Furthermore, the wavelength dependence of the scattering coefficient has been observed to follow
a power-law decay from the VIS-NIR to the SWIR region. Thus, the magnitude of tissue scattering is lower at
SWIR wavelengths than that observed at VIS or NIR wavelengths, potentially enabling increased penetration
depth of incident light at SWIR wavelengths that are not highly absorbed by the aforementioned chromophores.
These aspects of SWIR suggest that the tissue spectroscopy and imaging in this range of wavelengths have the
potential to provide enhanced sensitivity (relative to VIS-NIR measurements) to chromophores such as water
and lipids, thereby helping to characterize changes in the concentrations of these chromophores due to con-
ditions such as atherosclerotic plaque, breast cancer, and burns. ©2015 Society of Photo-Optical Instrumentation Engineers
(SPIE) [DOI: 10.1117/1.JBO.20.3.030901]
Keywords: short-wave infrared; near-infrared; tissue spectroscopy; multispectral imaging; tissue properties; optical properties;
absorption coefficient; scattering coefficient.
Paper 140807VR received Dec. 10, 2014; accepted for publication Feb. 24, 2015; published online Mar. 24, 2015.
1 Introduction
Spectroscopic and imaging techniques consisting of the delivery
of light to a biological tissue sample and detection of the dif-
fusely reflected or transmitted light from the sample are well-
established means of interrogating tissue composition, structure,
and function.1,2Most tissue optics research to date has employed
wavelengths in the visible and near-infrared regions of the spec-
trum (VIS-NIR, defined here as 400 to 1000 nm). A primary
goal of many VIS-NIR studies is to characterize the functional
status of a tissue by measuring the concentration of oxygenated
hemoglobin [primary absorption peak near 414 nm and secon-
dary absorption peaks near 543 and 577 nm (Ref. 3)] and deoxy-
genated hemoglobin [primary absorption peak near 433 nm and
secondary absorption peak near 556 nm (Ref. 3)]. In addition,
some studies have employed signals detected at the long wave-
length edge of the NIR (900 to 1000 nm) to extract parameters
related to tissue water and lipid concentrations, because water
and lipid have small absorption features near 970 (Ref. 4)
and 930 nm,5respectively. These studies have provided infor-
mation about changes in water fraction due to edema in
burns6and changes in relative amounts of hemoglobin, water,
and lipid content in breast tumors.1However, the VIS-NIR
region (as defined above) does not include many of the promi-
nent absorption peaks of water and lipids.
To obtain additional quantitative information about biologi-
cal tissue constituents, it may be advantageous to extend optical
measurements into the short-wave IR (SWIR) spectral region
(defined here as 1000 to 2000 nm). The SWIR regime includes
prominent absorption peaks of water, lipids,4,5,7and collagen8
(Fig. 1).
For lipids, the absorption peaks at 920, 1040, 1210, 1730,
and 1760 nm are associated with overtones of the stretching
vibrational mode of the C-H bond.9,10 The 920 and 1210 nm
peaks are associated with the second overtone of C-H stretch-
ing,10,11 while the 1730 and 1760 nm peaks are associated with
the first overtone of C-H stretching.1012 The absorption peak
near 1430 nm can be attributed to the first overtone of O-H
stretching.10,12 Different types of lipids, such as cholesterol,
cholesterol esters, phospholipids, and triglycerides, are known
to have absorption spectra that are similar to each other, with
distinctive peaks in the SWIR region.9
For collagen, the absorption spectrum has a peak at 1200 nm
from the second overtone of C-H stretching,11 a peak near
1500 nm from a combination band of CH2stretching and non-
stretching,13 a peak at 1725 nm due to symmetric and asymmet-
ric stretching bands from the CH2bond,11 and a shoulder at
1690 nm from the first overtone of CH3stretching.11
For water, the absorption peaks at 970 and 1180 nm are from
a vibrational overtone of the O-H bond.9The peak near 1430 nm
is due to the first overtone of O-H stretching,10,12 the peak near
1930 nm is due to an O-H stretch/deformation combination,10,12
and the peak near 1975 nm is due to an O-H bend second
overtone.10
*Address all correspondence to: Anthony J. Durkin, E-mail: adurkin@uci.edu
Journal of Biomedical Optics 030901-1 March 2015 Vol. 20(3)
Journal of Biomedical Optics 20(3), 030901 (March 2015) REVIEW
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/02/2015 Terms of Use: http://spiedl.org/terms
The absorption coefficient of water is over 60 times greater at
1440 nm than at the 970-nm NIR peak and over 260 times
greater at 1940 nm than at the 970-nm NIR peak.4Similarly,
the absorption coefficient of lipid is over 12 times greater at
1210 nm than at the 930-nm NIR peak, over 8 times greater at
1390 nm than at the 930-nm NIR peak, and over 80 times greater
at 1730 nm than at the 930-nm NIR peak.7Absorption peaks
from cholesterol are seen near 1200, 1400, and 1750 nm,14 sug-
gesting that the SWIR wavelength range has the potential to
capture information about cholesterol that is unavailable in the
VIS-NIR. Therefore, measurements at SWIR wavelengths could
provide improved sensitivity to the water and lipid content of
biological tissues. This increased sensitivity is likely to be
important for applications such as assessment and monitoring
of burns (which are marked by changes in water fraction due to
edema), characterization of atherosclerotic plaque (which can be
classified according to changes in lipid content), and detection
and monitoring of cancer (which is known to cause changes in
the structural and biochemical contents of tissue). Specific stud-
ies related to the aforementioned applications will be discussed
in the body of this paper.
Here, we present a review of SWIR techniques that have
been employed to characterize biological tissues. Early SWIR
studies largely consisted of reflectance and transmittance
measurements on thin slices of ex vivo tissue, using integrating
sphere setups.1520 More recently, SWIR measurements have
been performed with fiber-probe-based setups,2128 which
have the portability and flexibility required to enable wave-
length-resolved and time-resolved measurements of in vivo tis-
sues. Most clinically compatible SWIR tissue measurements
have only been tested on ex vivo samples, but some groups9,29,30
have begun to perform in vivo SWIR imaging of biological tis-
sues. These initial studies demonstrate that SWIR measurements
of tissue can provide a more spectral-information-rich dataset
for quantitative characterization of in vivo tissue structure and
function, including increased spectral information content about
water and lipid absorptions that are not available in the VIS-NIR
region. We will summarize several of these studies in the follow-
ing sections.
2 Integrating Sphere-Based Short-Wave
Infrared Measurements
Several studies have measured ex vivo tissue absorption [μaðλÞ]
and reduced scattering [μ0
sðλÞ] coefficients in the SWIR region
using integrating sphere setups.1520 Integrating sphere methods
have been widely employed in tissue optics to measure tissue
optical property spectra (e.g., Ref. 31). These methods involve
two measurements: a diffuse transmittance measurement, where
the sample is placed at the entrance port of the sphere, and
a diffuse reflectance measurement, where the sample is located
at the exit port at the back of the sphere. An unscattered (colli-
mated) transmittance measurement can also be performed, if
desired, to separate the reduced scattering coefficient into the
scattering coefficient [μsðλÞ] and the anisotropy factor [gðλÞ].
In the studies that we review here, light collected by the inte-
grating sphere setup was sent to detectors that are sensitive to
SWIR wavelengths [e.g., indium gallium arsenide (InGaAs)].
The reflectance [RðλÞ] and transmittance [TðλÞ] measurements
are then used in combination with a model of light propagation,
such as inverse adding doubling,31 to enable the determination
of the tissue absorption and reduced scattering coefficients.
Du et al.15 used an integrating sphere setup employing a tun-
able light source consisting of a tungsten lamp and a monochro-
mator to measure the optical properties of ex vivo porcine dermis
in the 900- to 1500-nm range, as skin optical properties had not
been well characterized in this range prior to this study. The
measured reflectance and transmittance spectra both exhibited
prominent dips in the 1400 to 1500 nm ranges, attributable
to an overtone of the O-H stretch vibrational mode of water,32
a result that corresponded with the trend in the reported
μaðλÞ, which was over 10 times greater at 1450 nm than at
980 nm.
Zamora-Rojas et al.16 used a double integrating sphere setup
to measure ex vivo porcine skin from 1150 to 2250 nm to inves-
tigate metrics for pork quality assessment. These measurements
employed a supercontinuum laser (460 to 2400 nm), a mono-
chromator (450 to 2800 nm), and InGaAs detectors.17 The
reflectance, transmittance, and collimated transmittance all
exhibited local minima near 1450 and 1950 nm. The value of
μaðλÞreached a local maximum of 3mm
1near 1450 nm,
but the values of μaðλÞfrom 1880 to 2040 nm could not be deter-
mined because high absorption within the samples resulted in
measured values of transmittance, TðλÞ, to be near zero in this
range. Overall, the absorption spectra of the skin tissue were
found to be primarily composed of features from water and
Fig. 1 (a) Absorption coefficients of oxygenated and deoxygenated
hemoglobin, water, and lipid in the visible and near-infrared (VIS-
NIR) (defined here as 400 to 1000 nm) and short-wave infrared
(SWIR) (defined here as 1000 to 2000 nm) regions, obtained
from Refs. 3,4, and 7, with each spectrum normalized to its maximum
value for ease of comparison. Both water and lipid have prominent
absorption peaks in the SWIR, despite not having many notable fea-
tures in the VIS or NIR regions. These spectra suggest that SWIR
measurements have the potential to provide tissue composition
information that is not nearly as readily available in the VIS-NIR.
(b) Absorption coefficients, from Ref. 8, of the same tissue constitu-
ents as in (a), from 500 to 1600 nm, with collagen and beta-carotene
added, with each spectrum normalized to its maximum value (repro-
duced with permission). Collagen exhibits a major SWIR absorption
peak near 1500 nm as well as secondary absorption peaks near 1050
and 1200 nm; most of this information content is not accessible with
VIS-NIR measurements.
Journal of Biomedical Optics 030901-2 March 2015 Vol. 20(3)
Wilson et al.: Review of short-wave infrared spectroscopy and imaging methods.. .
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/02/2015 Terms of Use: http://spiedl.org/terms
the scattering coefficient was found to decrease with increasing
wavelength.
Troy and Thennadil18 used a double integrating sphere setup
with a tungsten-halogen light source and InGaAs detectors to
measure human skin (22 samples, 14 subjects) from 1000 to
2200 nm. The study was designed to characterize the optical
properties of skin across a wide wavelength range, in order
to ultimately determine which window(s) of this range would
be most suitable for the use of a noninvasive glucose detection
instrument and to develop a quantitative method to correct glu-
cose measurements for the effects of tissue absorption and scat-
tering. The measured μaðλÞwas in agreement with a theoretical
calculation for 70% water. However, data could not be obtained
in the range of 1900 to 2040 nm due to high water absorption.
The measured μ0
sðλÞwas modeled with a power law of the form
μ0
sðλÞ¼ð2×104mm1Þðλλ0Þ1.5 , where λ0¼1nm. Over the
1000- to 2200-nm window, μ0
sðλÞvaried from 3 to 16 cm1.
In addition, the μ0
sðλÞvalue from 1000 to 1800 nm was
found to correlate inversely with age, a result potentially attrib-
utable to age-related decreases in cell density. Overall, the
study succeeded in quantifying the absorption and scattering
of human skin in the SWIR regime and determined that a
2% Intralipid solution could serve as an effective tissue simu-
lating phantom from 1000 to 2200 nm. These results can be
employed in future studies to motivate instrumentation and algo-
rithm design for accurate, noninvasive SWIR-based in vivo glu-
cose measurements.
Bashkatov et al.19 used an integrating sphere setup with a
halogen lamp and spectrometer for ex vivo measurements of
human tissues (skin, subcutaneous adipose, and mucosal) from
400 to 2000 nm. The measurement of subcutaneous adipose tis-
sue was motivated by the goal of developing optical techniques
for treating conditions such as obesity. The measurements of
skin and mucosal tissue were motivated by the need for
improved dosimetry during photodynamic therapy of disease
(using 850, 980, and 1064 nm light sources33) in tissues such as
the bladder, colon, esophagus, and maxillary sinus. The absorp-
tion spectra of the tissues had peaks near 1200, 1430, 1750, and
1925 nm [Fig. 2(a)], attributed to water and lipids. The μ0
sðλÞof
skin [Fig. 2(b)] was modeled as a sum of two power-law terms:
μ0
sðλÞ¼ð1.1 ×1011 Þλ4þð7.37Þλ0.22 mm1, with the λ4
term modeling Rayleigh scattering and the λ0.22 term modeling
Mie scattering. The reduced scattering coefficients of mucosal
and subcutaneous adipose tissues were approximated with sin-
gle-power-law models. A key conclusion of this study was that
the power-law behavior of μ0
sðλÞextended into the SWIR wave-
length range, demonstrating that the tissue scattering continues
to decrease monotonically in this regime.
Friebel et al.20 used an integrating sphere setup with a cuvette
to measure the optical properties of human red blood cells in
saline in the range 250 to 2000 nm. The study aimed to char-
acterize the optical properties of this suspension under different
oxygenation levels in order to determine how μaðλÞ, and pos-
sibly μsðλÞand gðλÞ, are influenced by oxygenation. The mea-
sured μaðλÞwas found to contain few features of hemoglobin at
wavelengths longer than 1100 nm. This result is consistent with
the fact that the oxy-hemoglobin absorption decreases by a
factor of 50 at 1000 nm, relative to its 576-nm absorption
peak, and that the deoxy-hemoglobin absorption decreases by
a factor of 250 at 1000 nm, relative to its 556 nm absorption
peak.3The measured μsðλÞfollowed a power-law model with an
exponent (b)of0.93 for wavelengths in the NIR and SWIR
regions. The measured gðλÞwas nearly flat (0.96 to 0.98)
from 1000 to 1750 nm, but it was observed to dip below 0.9
at 1900 nm, the location of the most prominent SWIR water
absorption peak. The μ0
sðλÞspectrum of the suspension exhibited
a corresponding peak at 1900 nm (an increase by roughly a fac-
tor of 2, relative to its value at 1750 nm). However, it is possible
that these effects may be artifacts of crosstalk between absorp-
tion and scattering in the data analysis procedure.
Anderson et al.34 used an integrating sphere spectrophotom-
eter setup to measure the absorption spectra of human sub-
cutaneous fat. A goal of these measurements was to identify
absorption peaks of lipid in the SWIR regime for use in selec-
tive photothermolysis.35 Prominent fat absorption peaks were
observed at 1210 and 1720 nm; at both of these wavelengths,
Fig. 2 (a) Absorption spectrum μaðλÞfrom 400 to 2000 nm for human
skin tissue (N¼21 samples) measured ex vivo in Ref. 19, with promi-
nent absorption peaks labeled. (b) Reduced scattering spectrum μ0
sðλÞ
from 400 to 2000 nm for the same human skin tissue dataset as in (a),
modeled as a sum of Rayleigh scattering (λ4, most prominent at vis-
ible wavelengths) and Mie scattering (λ0.22, most prominent at NIR
and SWIR wavelengths). Error bars represent standard deviation
over the number of samples. Notable water and lipid absorption
peaks are present throughout a large portion of the SWIR region,
and the reduced scattering spectrum can be approximated by a
Mie scattering power law in the SWIR region. These spectra suggest
that SWIR measurements can be employed to detect spectral signa-
tures of water and lipid in skin and that the power-law behavior of the
reduced scattering coefficient extends into the SWIR regime. (Figure
© Institute of Physics and Engineering in Medicine. Reproduced with
permission of IOP publishing. All rights reserved).
Journal of Biomedical Optics 030901-3 March 2015 Vol. 20(3)
Wilson et al.: Review of short-wave infrared spectroscopy and imaging methods.. .
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/02/2015 Terms of Use: http://spiedl.org/terms
the absorption from fat exceeded the absorption of water. This
result suggests that the lipids have notable spectral features in
the SWIR regime that can be selectively targeted due to their
spectral separation from the prominent absorption peaks of
water.
Martin36 examined the SWIR absorption of human skin in
vivo by placing a detector, comprised of a spectrograph and
an integrating sphere, directly onto the tissue surface. The detec-
tor measured reflectance from 1200 to 2400 nm; prominent skin
absorption peaks near 1450, 1775, and 1900 nm corresponded to
the absorption peaks of water in those spectral regions. Minima
near 1875, 1890, 1910, and 1925 nm in the second derivative of
the skin reflectance spectrum were attributed to different binding
states of water, and the relative intensities of the reflectance sig-
nals at these bands were observed to change in response to
occlusion and variation in relative humidity.
The main conclusions of the SWIR integrating sphere studies
were: (1) the tissue scattering coefficient obeyed a power law
over the SWIR wavelength region and the NIR wavelength
region, and (2) tissue absorption coefficients showed prominent
water- and lipid-related absorption features in the SWIR region.
These measurements highlighted prominent spectral features of
tissue absorption and scattering in the SWIR regime; however,
the integrating sphere method is fundamentally limited to ex
vivo tissue samples that are of appropriate thickness to enable
collection of both diffuse reflectance and diffuse transmittance.
3 Fiber-Probe-Based Short-Wave Infrared
Measurements
SWIR measurements have also been performed using point
probe geometries that employ optical fibers2128 to obtain mea-
surements of reflectance collected over small localized tissue
volumes that are adjacent (within microns to millimeters) to
the point of contact. The interrogation volume of these probes
is defined in part by the spacing between the source and detector
fibers.37 Fiber-probe-based methods enable interrogation of ex
vivo and in vivo intact tissues because, unlike integrating sphere
setups, these methods typically employ measurement geom-
etries that can be approximated as semi-infinite by radiative
transport modeling approaches and thus do not require concur-
rent reflectance and transmittance measurements or an a priori
knowledge of sample thickness. These fiber-based systems can
also be employed preclinically or clinically due to their portable,
flexible, and minimally invasive nature. Fiber-based methods
have shown that the use of steady-state and time-resolved SWIR
information contents related to water and lipids can enable
improved detection of diseases such as atherosclerosis and
cancer, as will be reviewed below. One of these groups21 directly
showed that adding SWIR information content to VIS-NIR
information content can provide improved confidence levels for
cancer detection (relative to using VIS-NIR alone). However, it
is important to remember that the sampled volume of tissue is a
function of the source-detector separation of the fiber probe.
Therefore, for layered tissues (such as skin), the reflectance
spectra will be influenced by a combination of the optical prop-
erties of the different tissue layers, weighted according to the
percentage of photon path spent in each layer (which is in
part defined by the source-detector separation).
Within the context of cardiovascular disease detection, there
have been a number of investigations that suggest the potential
of SWIR-derived information to enhance understanding related
to pathologic changes in tissue composition. Wang et al.22 used a
fiber probe and spectrometer to measure the absorbance [defined
as log(1/reflectance)] of ex vivo carotid atherosclerotic plaques
obtained from 25 patients (Fig. 3). Lipid-to-protein ratios in
the spectral regions from 1130 to 1260 nm, 1620 to 1820 nm,
and 2200 to 2330 nm were employed to distinguish between
vulnerable, stable, and intermediate plaques. Moreno et al.23
performed ex vivo reflectance measurements from 1100 to
2200 nm on 199 human aorta samples obtained at autopsy.
Multivariate statistical analysis was carried out to classify the
spectra, and principal components of these data were employed
for differentiating stable plaque from vulnerable plaque.
Nachabé et al.24 employed a fiber probe to measure ex vivo
porcine tissues (fat, muscle, and white matter) from 900 to
1600 nm to validate a noninvasive, clinically compatible method
for extracting water and lipid volume fractions from tissue. In
a related study, Nachabé et al.21 directly compared two fiber-
probe-based optical spectroscopic methods for obtaining tissue
water and lipid volume fractions: one method employing only
VIS-NIR wavelengths (500 to 1000 nm) and other method
including SWIR wavelengths in addition to VIS-NIR (500 to
1600 nm). These methods involved both measuring the reflec-
tance spectrum of the tissue and modeling the reflectance spec-
trum as a function of the wavelength-dependent absorption and
scattering coefficients of the tissue. The absorption coefficient
was represented as a linear combination of all the absorbing
species in the tissue, weighted according to their respective vol-
ume fractions. These two methods were employed to measure
ex vivo porcine fat and muscle from 500 to 1600 nm (with the
VIS-NIR-SWIR technique using silicon and InGaAs spectrom-
eters in tandem).21 The confidence levels associated with the
water and lipid volume fractions obtained over the entire 500-
to 1600-nm range were four times higher than those obtained
from only 500 to 1000 nm.21
The results of Ref. 21 were then used by Nachabé et al.8to
develop a breast cancer detection method employing optical
spectroscopic measurements of ex vivo human breast tissues
from 500 to 1600 nm (Fig. 4) to distinguish between adipose
tissue, glandular tissue, fibroadenoma, invasive carcinoma,
and ductal carcinoma in situ (DCIS). The water volume fraction
was statistically significant for distinguishing fibroadenoma
Fig. 3 SWIR absorbance spectra [unitless, defined as 1/log(reflec-
tance)] of human carotid atherosclerotic plaques (fibrous, dotted
line; calcified, dashed line; and soft, solid line), measured ex vivo
using a fiber probe and spectrometer.22 The spectra exhibit features
of absorption from water (peaks near 1450 and 1950 nm) and lipids
(peak near 1200 nm). This work suggests that SWIR may have the
potential to differentiate between vulnerable plaque and more stable
forms of plaque. (Figure reproduced with permission.)
Journal of Biomedical Optics 030901-4 March 2015 Vol. 20(3)
Wilson et al.: Review of short-wave infrared spectroscopy and imaging methods.. .
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/02/2015 Terms of Use: http://spiedl.org/terms
from adipose tissue, distinguishing invasive carcinoma from adi-
pose, glandular, and fibroadenoma, and distinguishing DCIS
from adipose tissue. The lipid volume fraction was statistically
significant for distinguishing fibroadenoma from adipose tissue,
distinguishing invasive carcinoma from adipose tissue, and
distinguishing DCIS from adipose, fibroadenoma, and invasive
carcinoma. The collagen volume fraction was statistically sig-
nificant for distinguishing fibroadenoma from glandular tissue,
distinguishing invasive carcinoma from adipose and fibroade-
noma, and distinguishing DCIS from adipose tissue.
Olesberg et al.25 employed a fiber-probe-based method to
measure the absorbance of rat skin in vivo from 2000 to 2500 nm
to optically measure glucose concentration. A tungsten-halogen
light source coupled to a fiber probe with a ball lens was used for
delivery of light to the skin. Photons that were transmitted
through a fold in the animals skin were detected by a second
fiber probe, which sent the light to an InGaAs detector and a
spectrometer. The absorption spectra were analyzed with a par-
tial least squares model to determine spectral features that were
related to glucose. This model was then employed to extract
the change in glucose concentration in the skin over a 2-h
period following venous delivery of glucose to the animal. The
extracted magnitude (5mMat baseline to 35-mM peak con-
centration following infusion) and temporal lineshape of the glu-
cose concentration were found to be in good agreement (within
15% at peak) with that obtained using a glucose analyzer.
Nishimura et al.26 employed a time-resolved fiber-probe-
based setup to measure absorption and reduced scattering coef-
ficients from 1150 to 1520 nm in a human forearm in vivo
(Fig. 5). The model fit to the data, using a water fraction in
the neighborhood of 52% (a value in good agreement with
that obtained using magnetic resonance imaging), was accurate
from 1150 to 1350 nm. However, the fit was notably worse at
wavelengths of 1400 nm and longer; to explain this discrepancy,
the authors suggest that their diffusion-based model was likely
inaccurate in this region, partly because absorption from water is
so high in this range that the photon paths cannot be considered
to be diffuse. In a proof-of-principle study, Bargigia et al.27 used
a fiber probe to obtain time-resolved spectroscopic measure-
ments of porcine fat ex vivo from 1100 to 1700 nm, observing
1200-, 1400-, and 1700-nm lipid absorption peaks and a reduced
scattering coefficient with values in the 1 to 4cm
1range from
1100 to 1700 nm. Taroni et al.28 obtained the time-resolved
in vivo reflectance of human breast tissue from 900 to 1300 nm,
reporting an absorption peak around 1200 nm containing con-
tributions from water, lipids, and collagen.
The main conclusions of the fiber-probe-based SWIR studies
were: (1) fiber-probe configurations enable SWIR measure-
ments of intact biological tissues, both in vivo and ex vivo,
(2) adding the SWIR range to these measurements enables
more accurate extraction21 of water and lipid volume fractions
than similar measurements over only the VIS-NIR range, and
(3) the water and lipid content obtained from fiber-based
SWIR measurements show some promises in enabling accurate
classification of diseased tissues; for instance, distinguishing
vulnerable atherosclerotic plaque from stable plaque22,23 and
distinguishing different types of breast cancer from nonmalig-
nant breast tissues.8
4 Short-Wave Infrared Imaging
SWIR methods have recently been employed for imaging appli-
cations.9,29,30 Imaging techniques provide several advantages
over fiber-probe-based methods; specifically, they enable non-
contact measurement of in vivo or ex vivo tissues and provide
information from fields of view that may be up to tens of
centimeters in both the x- and y-dimensions.
Allen et al.9performed ex vivo spectroscopic photoacoustic
imaging of human aorta with plaques from 740 to 1400 nm, also
imaging a plaque buried beneath 2.8 mm of blood to demon-
strate in vivo feasibility. The authors of Ref. 9cited SWIR
Fig. 4 Diffuse reflectance spectra (colored symbols) of (a)(c) noncancerous and (d) and (e) cancerous
human breast tissues over the range from 500 to 1600 nm, measured ex vivo with a fiber-probe-based
setup and shown with fits of a diffusion theory model (black solid lines).8Differences between tissue types
can be clearly seen and quantitatively related to changes in water and lipid volume fractions, which can
then be employed for tissue classification. (Figure reproduced with permission.)
Journal of Biomedical Optics 030901-5 March 2015 Vol. 20(3)
Wilson et al.: Review of short-wave infrared spectroscopy and imaging methods.. .
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/02/2015 Terms of Use: http://spiedl.org/terms
photoacoustic imaging as a potential means of providing
improved diagnosis of vulnerable plaque, due to its ability to
interrogate structure and function of tissue with a high sensitiv-
ity to lipid absorption features in the SWIR wavelength range.
Fleming et al.14 performed spectroscopic optical coherence
tomography (OCT) with a system designed for use through a
catheter,38 to image tissue phantoms with lipid emulsion inclu-
sions (representing coronary plaques) injected 1mmbeneath
the surface of porcine aortic tissue. The center wavelength and
bandwidth of the OCT system were 1300 and 100 nm, respec-
tively; therefore, the system acquired data from 1250 to
1350 nm. The spectral data and OCT images were employed
to generate probability maps for collagen and cholesterol
beneath the surface of the phantom. The plaques were detected
as regions with a higher probability of cholesterol and a lower
probability of collagen, relative to the surrounding medium. The
cholesterol inclusion was clearly visible in a region beneath
0.79 mm of aortic tissue, but it was not visible beneath 1.17 mm
of tissue. This study suggests that the multispectral SWIR im-
aging has the potential to distinguish spectral signatures of cho-
lesterol and collagen for diagnosis of arterial plaque.
To test the ability of SWIR wavelengths to noninvasively
interrogate biological tissues, Cao et al.29 measured transmitted
intensity of ex vivo murine skeletal muscle, liver, kidney, cardiac
tissue, cerebrum, cerebellum, and adipose tissues as a function
of wavelength from 800 to 1450 nm. Cao et al.29 also performed
SWIR transmittance imaging from 800 to 1400 nm through the
head of a mouse in vivo, using the ratio of measured intensities
at 1075 and 975 nm as a contrast metric, to demonstrate the
potential of NIR/SWIR methods to provide an image of the
skull that could be coregistered, in terms of gross anatomy, with
images that were obtained by x-ray.
Akbari et al.39 used a hyperspectral VIS-NIR-SWIR imaging
system for detection and monitoring of intestinal ischemia in
vivo in a porcine model. The instrumentation included two hal-
ogen lamps for illumination and two cameras (a 400- to 1000-
nm camera and a 900- to 1700-nm camera) for image detection.
The SWIR (900 to 1700 nm) camera provided a spectral reso-
lution of 5 nm by use of a 30-m slit. The slit provided one thin
sliceof the ðx; y; λÞimage at a time, so the slit was scanned
across the object using a push-broommethod to obtain the
entire wide-field image. Images were obtained for 157 wave-
length bands in the 900- to 1700-nm range; each image took
15 s to acquire and the spatial resolution of the images was
0.35 mmpixels. In the ischemia experiment, a clamping
procedure was performed on a portion of the intestine and the
associated blood vessels. Wide-field reflectance images of the
exposed abdominal cavity, with peritoneum, spleen, intestine,
and bladder within the field of view, were acquired intraopera-
tively before clamping and 2,4, and 6 min after clamping.
The reflectance data were analyzed with a quantitative, feature-
based method, in order to locate spectral regions in which the
lineshape measured from ischemic tissue was the most notably
different than that measured from normal tissue. To accomplish
this, the derivative of the reflectance spectrum was evaluated at
each wavelength, and the absolute value of the sum of these
derivatives was calculated for different subsets of the measured
wavelength range. It was found that the range from 999 to
1182 nm was optimal for distinguishing ischemic spectra from
normal spectra, so the magnitude of the sum of the derivatives in
this region was defined to be the ischemia index. Camera pixels
with ischemia index values that surpassed a specific threshold
were labeled as ischemic. Using the SWIR images, the regions
of ischemia were classified with a false positive rate of 8% and a
false negative rate of 7%. Akbari et al.40 also performed a similar
study on freshly excised human gastric tumors, using the same
technology as in Ref. 39. With a mathematical approach similar
to that employed in Ref. 39,acancer indexwas calculated by
summing the square of the reflectance derivative from 1226 to
1251 nm and the square of the reflectance derivative from
1288 to 1370 nm. Despite being based purely on mathematical
differences in spectral lineshape instead of biophysical tissue
optics parameters, the model distinguished cancerous tissues
from noncancerous tissues with a false positive rate of 7% and
a false negative rate of 9%. Furthermore, the method showed
potential for detecting tumors buried beneath up to 2to
3 mm of benign mucosal tissue.
Randeberg et al.41 employed an SWIR imaging system to
assess human skin bruises. The instrument acquired images
in 160 spectral bands from 900 to 1700 nm. The study
Fig. 5 (a) Effective attenuation, μeffðλÞand (b) absorption, μaðλÞcoefficients of a human forearm in vivo,
from 1150 to 1520 nm, obtained from time-resolved reflectance with a fiber probe.26 The measured μa
values in (b) are shown alongside models for μaðλÞusing water fractions of 47%, 52%, and 57%, rep-
resented with the broken line, solid line, and dotted line, respectively, employing the water absorption
spectrum from Ref. 4. The water fraction of 52% was in good agreement with that obtained for muscle
tissue using magnetic resonance imaging. This result suggests that the time-resolved SWIR imaging has
the potential to quantitatively and noninvasively sense water content beneath the tissue surface in vivo.
(Figure © Institute of Physics and Engineering in Medicine. Reproduced with permission of IOP publish-
ing. All rights reserved).
Journal of Biomedical Optics 030901-6 March 2015 Vol. 20(3)
Wilson et al.: Review of short-wave infrared spectroscopy and imaging methods.. .
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/02/2015 Terms of Use: http://spiedl.org/terms
determined that the 958-nm image was best for distinguishing
vasculature, but the 1584-nm image was best for distinguishing
the bruise from the surrounding tissue. In a subsequent study,
Randeberg and Hernandez-Palacios42 employed an SWIR
imager that spanned the 950- to 2500-nm range with 256 spec-
tral bands and 6-nm spectral sampling to image and characterize
skin bruises. A minimum noise transform was employed to
enhance the qualitative features of the measured SWIR spectra.
The images at 1250, 1443, and 2165 nm were employed to
construct false-color RGBmaps of the bruises. The SWIR
absorption contrast (attributed to increased water concentration
in the bruise due to edema) was found to provide better distinc-
tion of bruised tissue than the VIS-NIR absorption contrast
(attributed to local changes in blood content). These results sug-
gest that SWIR imaging could provide notable improvement in
characterization of bruises relative to visual inspection of the
color of the bruise, which is the current clinical gold standard
for bruise assessment.4144
Recently, in vivo wide-field SWIR imaging of burns in a rat
model was performed using a hybrid approach involving both
spatially modulated and unstructured (homogeneous) illumina-
tions.30 This method enabled the separation of absorption and
reduced scattering coefficients over a wide (850 to 1800 nm)
wavelength range (Fig. 6), showing potential to distinguish
burn-related changes in tissue absorption and scattering in the
SWIR regime. The measured tissue absorption and reduced scat-
tering coefficients obtained from this study were also input into
a Monte Carlo model45 to obtain reflectance and transmittance
spectra from 850 to 1800 nm (Fig. 7). The reflectance and trans-
mittance curves in Fig. 7illustrate that, in spite of the decreased
scattering in the SWIR regime relative to the NIR, there are still
some SWIR wavelengths (most notably from 1400 to 1500 nm)
that have very low transmittance due to the high absorption
coefficient of water in that range of wavelengths.
The main conclusions of the SWIR imaging studies were:
(1) the SWIR regime provides sufficient optical penetration
depth (e.g., up to 5mmfor 1100 nm photons, using the equa-
tion employed in Ref. 19) to interrogate subsurface tissue fea-
tures, such as changes in vasculature and collagen structure in
burns and cancerous tissues, in a noninvasive manner over a
wide ðtens of centimetersÞ2field of view, and (2) the enhanced
absorption contrast from water and lipid in the SWIR regime
(relative to the VIS-NIR) provides the potential for quantitative
wide-field mapping of changes in tissue chromophore concen-
trations in perturbed (e.g., burned and diseased) tissues.
5 Discussion and Conclusions
SWIR measurements of biological tissues are capable of
detecting changes in absorption from chromophores such as
water [absorption peaks at 1150, 1450, 1900 nm (Refs. 4and
8)] and lipids [absorption peaks at 930, 1040, 1200, 1400,
and 1700 nm (Refs. 5,7, and 8)], as well as possibly collagen
[absorption peaks at 1200 and 1500 nm (Ref. 8)]. For each of
these tissue constituents, prominent spectral characteristics like
those observed in the SWIR regime are not present in the VIS-
NIR, suggesting that SWIR spectroscopy and imaging tech-
niques provide potential for enabling a more information-rich
means of tissue diagnostics and characterization.
Thus far, many SWIR tissue optics studies might generally
be described as exploratory and proof-of-principlein nature.
The current state-of-the-art has primarily demonstrated that
(1) SWIR tissue optics techniques have been employed to access
spectral bands that are generally related to water and lipid
content, and (2) employing the SWIR enables extraction of
an expanded set of tissue parameters that may be useful for dis-
ease detection, classification, characterization, and monitoring.
Additional research needs to be performed in order to quantita-
tively assess the value of expanding existing instrumentation
into the SWIR range in terms of improved diagnostic capability.
One of the reasons that SWIR has seen relatively limited
application in the biomedical sector up until recently is that
InGaAs sensor arrays,46 which are often used for SWIR detec-
tion, have been cost prohibitive. However, recent technological
advances, coupled with an increase in the number of companies
supplying this technology, have enabled a drop in price of
roughly an order of magnitude over the time period from
2010 to 2014. In addition, access to high-resolution InGaAs sen-
sor arrays has been limited by various national defense-related
policies such as International Traffic in Arms Regulations. An
increased supply of affordable SWIR sensing instrumentation
has the potential to enable more widespread SWIR-oriented bio-
medical optics research and development.
Fig. 6 Reduced scattering (top panel) and absorption (three bottom
panels) coefficients from 850 to 1800 nm obtained for in vivo rat tissue
preburn and 2-h postburn using a hybrid wide-field imaging method
with spatially modulated and unmodulated light.30 (Figure reproduced
with permission).
Journal of Biomedical Optics 030901-7 March 2015 Vol. 20(3)
Wilson et al.: Review of short-wave infrared spectroscopy and imaging methods.. .
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/02/2015 Terms of Use: http://spiedl.org/terms
Two examples of application areas in which there has been
notable activity with a potential to exploit the SWIR regime are
the identification of vulnerable atherosclerotic plaque and clas-
sification of stages of malignancy in breast cancer.
SWIR applications related to atherosclerosis show specific
promise because several studies have shown that the lipid
absorption information in this range can potentially distinguish
vulnerable plaque.9,14,22,23 For eventual in vivo translation, the
SWIR provides an advantage over the VIS-NIR for detecting
changes in lipids amidst a high concentration of blood (which
absorbs much less prominently in the SWIR than in the VIS-
NIR). In fact, one group has already employed these concepts
to design a clinically compatible instrument for plaque detection
that accesses the SWIR portion of the spectrum.4749 This
example has been successfully translated into a commercial
product, the catheter-based TVC Imaging System developed
by Infraredx®, for determining information about the structure
and plaque content of blood vessels in vivo.50
SWIR applications related to breast cancer classification are
also particularly promising because previous NIR studies have
already shown that water and lipid content are of great potential
use in distinguishing breast cancer,51,52 and another study has
shown that including SWIR wavelengths can enable extraction
of water and lipid volume fractions with a factor of four increase
in confidence level relative to those obtained using only VIS-
NIR light, by employing the enhanced absorption contrast
from these tissue components in the SWIR.21
Additional biomedical applications of SWIR technology
include quantitative sensing of biophysical tissue property
changes in burn wounds (employing the increased sensitivity
of SWIR wavelengths to absorption features from water, enabling
detection of edema in burned tissue)30 and detection of changes in
glucose concentration (if the spectral range is extended out to
2500 nm).25 The applications described in this paper illustrate
the potential of SWIR tissue optical spectroscopy and imaging
to provide enhanced diagnostic information.
The scattering of biological tissue has been shown to follow
a power-law model over the VIS-NIR-SWIR regimes, and the
absorption of tissue has many distinctive spectral features in the
SWIR. Therefore, employing SWIR wavelengths can poten-
tially enable interrogation of many different photon penetration
depths within the tissue. Layered tissues, such as skin, have dif-
ferent absorption and scattering properties in each layer due to
the heterogeneity of tissue morphology and chromophore con-
centration. As a result, the ability to interrogate different depths
may be particularly helpful for characterizing layered tissues
such as skin.
Incorporating the SWIR wavelength region into tissue optics
measurements can potentially help to improve in vivo tissue
characterization. In the VIS-NIR regime, tissue absorption is
primarily attributed to hemoglobin, which exhibits strong spec-
tral peaks near 420, 540, and 577 nm. However, in the SWIR,
absorption from hemoglobin is multiple orders of magnitude
lower than in the VIS-NIR, while the SWIR absorption
peaks of water and lipid are typically at least an order of mag-
nitude greater than those at the far edge of the NIR. Therefore,
for in vivo optical spectroscopy and imaging studies, where tis-
sue is interrogated with a probe attached to an endoscope or a
catheter, the detected SWIR signal will be largely unaffected by
the presence of blood in the imaged field. In addition, the signal
will be much more sensitive to water and lipids than it would be
in the VIS-NIR. As a result, SWIR wavelengths may provide
valuable subsurface information about the water and lipid con-
tent of human tissues in vivo. This information can potentially
assist in the detection, characterization, and classification of
conditions such as cancer and atherosclerosis. In fact, it has
already been shown that water and lipid absorptions provide
useful diagnostic fingerprintsfor detection of both cancer8,40
and atherosclerosis.9,14,22,23,4749
On the path toward a more widespread implementation of
SWIR tissue optics measurements, one important factor to con-
sider is the temperature dependence of tissue absorption features
in the SWIR regime. Water is a primary SWIR absorber in tis-
sue, and it has been shown53,54 that the spectral location of
SWIR water absorption peaks can vary with temperature.
Jansen et al.53 reported a water absorption coefficient that
decreased with increasing temperature and a water absorption
peak that shifted from 1940 nm at 22°C to 1920 nm at 70°C.
Kakuta et al.54 reported a blue shift in the 1450 nm water absorp-
tion peak when the temperature was varied between 24°C and
40°C. Furthermore, temperature-dependent changes in the mea-
sured optical properties of skin have been reported in the VIS-
NIR,55,56 a finding that suggests that such changes are likely
expected in the SWIR as well. The results in Refs. 55 and
56 were cited by Troy and Thennadil18 as motivation for per-
forming all their SWIR integrating sphere measurements of
skin tissue at body temperature (37°C). Temperature-dependent
shifts in the SWIR tissue absorption peaks can potentially be
reduced or eliminated through careful control of temperature,
or if such control is not possible, the locations of the peaks
can potentially be reshifted by calibrating the SWIR instrumen-
tation through measurements of tissue constituents (e.g., water
and lipid) as a function of temperature and using this calibration
to correct the peak locations in a postprocessing routine.
In conclusion, the application of SWIR-based optical spec-
troscopy and imaging to in vivo tissue has, to date, been rela-
tively limited. Studies suggest that this wavelength range has
potential for enhancing the current state-of-the-art in tissue
optics measurements, which, at present, typically only employ
light in the VIS-NIR spectral range. This enhancement is
Fig. 7 Reflectance and transmittance of rat skin in the SWIR range,
calculated using Monte Carlo simulation.45 The absorption and
reduced scattering coefficients input into the simulations are those
shown in Fig. 6for normal rat tissue in vivo, as obtained from multi-
spectral SWIR imaging.30 These reflectance and transmittance curves
indicate that although scattering decreases monotonically over the
SWIR regime, the mean penetration depth of the light does not
increase monotonically as a result. Instead, the SWIR penetration
depth curve for skin exhibits a number of local maxima and minima
that coincide with the absorption features of water in this wavelength
range.
Journal of Biomedical Optics 030901-8 March 2015 Vol. 20(3)
Wilson et al.: Review of short-wave infrared spectroscopy and imaging methods.. .
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/02/2015 Terms of Use: http://spiedl.org/terms
primarily provided by increased sensitivity to endogenous tissue
components (such as water, lipids, and collagen) that absorb
light prominently in the SWIR but do not have strong absorption
in the VIS-NIR. Incorporating SWIR detection into biomedical
optics technologies has the potential to provide additional
information regarding tissue composition and function and
as such may be useful for enhanced tissue assessment and
classification.
Acknowledgments
R.H.W. was supported by a postdoctoral fellowship from the
Hewitt Foundation for Medical Research. K.P.N. was supported
by the NSF IGERT (Grant 1144901), Biophotonics across
Energy, Space, and Time.F.B.J. was internally funded by
Raytheon Vision Systems. Additional support is provided
by the National Institutes of Health (NIH-R21EB014440
and NIH-R21NS078634), the NIH/NIBIB funded LAMMP
(P41EB015890), the Military Medical Photonics Program
(FA9550-10-1-0538), and the Arnold and Mabel Beckman
Foundation.
References
1. T. D. OSullivan et al., Diffuse optical imaging using spatially and
temporally modulated light,J. Biomed. Opt. 17(7), 071311 (2012).
2. J. Q. Brown et al., Advances in quantitative UV-visible spectroscopy
for clinical and pre-clinical application in cancer,Curr. Opin.
Biotechnol. 20(1), 119131 (2009).
3. S. Prahl, Optical Absorption of Hemoglobin, Oregon Medical Laser
Center, Portland, OR, omlc.ogi.edu/spectra/hemoglobin (9 March 2015).
4. G. M. Hale and M. R. Querry, Optical constants of water in the 200 nm
to 200 μmwavelength region,Appl. Opt. 12(3), 555563 (1973).
5. R. L. P. van Veen et al., Determination of VIS-NIR absorption
coefficients of mammalian fat, with time and spatially resolved diffuse
reflectance and transmittance spectroscopy,in OSA BIOMED Topical
Meeting, Miami Beach, FL (2004).
6. J. Q. Nguyen et al., Spatial frequency domain imaging of burn wounds
in a preclinical model of graded burn severity,J. Biomed. Opt. 18(6),
066010 (2013).
7. G. B. Altshuler, R. R. Anderson, and D. Manstein, Method and appa-
ratus for the selective targeting of lipid-rich tissue,United States Patent,
Patent No. 6,605,080 B1 (2003), http://www.npsg.uwaterloo.ca/data/
skin/lipid_absorption.m (9 March 2015).
8. R. Nachabé et al., Diagnosis of breast cancer using diffuse optical spec-
troscopy from 500 to 1600 nm: comparison of classification methods,
J. Biomed. Opt. 16(8), 087010 (2011).
9. T. J. Allen et al., Spectroscopic photoacoustic imaging of lipid-rich
plaques in the human aorta in the 740 to 1400 nm wavelength
range,J. Biomed. Opt. 17(6), 061209 (2012).
10. H. M. Monavar et al., Evaluation of the Potential of Near-Infrared (NIR)
Spectroscopy for Predicting Fatty Acid Composition of SturgeonsCaviars,
SpectroscopySolutions, 2013, http://www.spectroscopy-solutions.org/
(9 March 2015).
11. P. Wang et al., Mapping lipid and collagen by multispectral photo-
acoustic imaging of chemical bond vibration,J. Biomed. Opt. 17(9),
096010 (2012).
12. S. S. Nielsen, Food Analysis, 4th ed., Springer, New York (2010).
13. P. Wang et al., Bond-selective imaging of deep tissue through the opti-
cal window between 1600 and 1850 nm,J. Biophotonics 5(1), 2532
(2012).
14. C. P. Fleming et al., Depth resolved detection of lipid using spectro-
scopic optical coherence tomography,Biomed. Opt. Express 4(8),
12691284 (2013).
15. Y. Du et al., Optical properties of porcine skin dermis between 900 nm
and 1500 nm,Phys. Med. Biol. 46(1), 167181 (2001).
16. E. Zamora-Rojas et al., Optical properties of pig skin epidermis and
dermis estimated with double integrating spheres measurements,
Innovative Food Sci. Emerging Technol. 20, 343349 (2013).
17. E. Zamora-Rojas et al., Double integrating sphere measurements for
estimating optical properties of pig subcutaneous adipose tissue,
Innovative Food Sci. Emerging Technol. 19, 218226 (2013).
18. T. L. Troy and S. N. Thennadil, Optical properties of human skin in the
near infrared wavelength range of 1000 to 2200 nm,J. Biomed. Opt.
6(2), 167176 (2001).
19. A. N. Bashkatov et al., Optical properties of human skin, subcutaneous
and mucous tissues in the wavelength range from 400 to 2000 nm,
J. Phys. D 38(15), 25432555 (2005).
20. M. Friebel et al., Influence of oxygen saturation on the optical scatter-
ing properties of human red blood cells in the spectral range 250 to
2,000 nm,J. Biomed. Opt. 14(3), 034001 (2009).
21. R. Nachabé et al., Estimation of biological chromophores using diffuse
optical spectroscopy: benefit of extending the UV-VIS wavelength
range to include 1000 to 1600 nm,Biomed. Opt. Express 1(5),
14321442 (2010).
22. J. Wang et al., Near-infrared spectroscopic characterization of human
advanced atherosclerotic plaques,J. Am. Coll. Cardiol. 39(8), 1305
1313 (2002).
23. P. R. Moreno et al., Detection of lipid pool, thin fibrous cap, and
inflammatory cells in human aortic atherosclerotic plaques by near-
infrared spectroscopy,Circulation 105(8), 923927 (2002).
24. R. Nachabé et al., Estimation of lipid and water concentrations in scat-
tering media with diffuse optical spectroscopy from 900 to 1,600 nm,
J. Biomed. Opt. 15(3), 037015 (2010).
25. J. T. Olesberg et al., In vivo near-infrared spectroscopy of rat skin tissue
with varying blood glucose levels,Anal. Chem. 78(1), 215223 (2006).
26. G. Nishimura, I. Kida, and M. Tamura, Characterization of optical
parameters with a human forearm at the region from 1.15 to 1.52 μm
using diffuse reflectance measurements,Phys. Med. Biol. 51(11),
29973011 (2006).
27. I. Bargigia et al., Time-resolved diffuse optical spectroscopy up to
1700 nm by means of a time-gated InGaAs/InP single-photon avalanche
diode,Appl. Spectrosc. 66(8), 944950 (2012).
28. P. Taroni et al., First in vivo spectral characterization of breast up to
1300 nm,Proc. SPIE 7896, 78962L (2011).
29. Q. Cao et al., Multispectral imaging in the extended near-infrared
window based on endogenous chromophores,J. Biomed. Opt. 18(10),
101318 (2013).
30. R. H. Wilson et al., Quantitative short-wave infrared multispectral im-
aging of in vivo tissue optical properties,J. Biomed. Opt. 19(8), 086011
(2014).
31. J. W. Pickering et al., Double-integrating-sphere system for measuring
the optical properties of tissue,Appl. Opt. 32(4), 399410 (1993).
32. R. A. Ryerson, Ed., Remote Sensing for the Earth Sciences, John Wiley
& Sons, New York (1999).
33. A. Roggan and G. Müller, Eds., Laser-Induced Interstitial Thermother-
apy, SPIE, Bellingham, Washington (1995).
34. R. R. Anderson et al., Selective photothermolysis of lipid-rich tissues:
a free electron laser study,Lasers Surg. Med. 38(10), 913919 (2006).
35. J. A. Parrish and R. R. Anderson, Selective photothermolysis: precise
microsurgery by selective absorption of pulsed radiation,Science
220(4596), 524527 (1983).
36. K. Martin, In vivo measurements of water in skin by near-infrared
reflectance,Appl. Spectrosc. 52(7), 10011007 (1998).
37. H. Arimoto, M. Egawa, and Y. Yamada, Depth profile of diffuse reflec-
tance near-infrared spectroscopy for measurement of water content in
skin,Skin Res. Technol. 11(1), 2735 (2005).
38. S. H. Yun et al., Comprehensive volumetric optical microscopy in
vivo,Nat. Med. 12(12), 14291433 (2006).
39. H. Akbari, K. Kojima, and N. Tanaka, Detection and analysis of the
intestinal ischemia using visible and invisible hyperspectral imaging,
IEEE Trans. Biomed. Eng. 57(8), 20112017 (2010).
40. H. Akbari et al., Cancer detection using infrared hyperspectral imag-
ing,Cancer Sci. 102(4), 852857 (2011).
41. L. L. Randeberg et al., Hyperspectral imaging of bruised skin,Proc.
SPIE 6078, 60780O (2006).
42. L. L. Randeberg and J. Hernandez-Palacios, Hyperspectral imaging of
bruises in the SWIR spectral region,Proc. SPIE 8207, 82070N (2012).
43. N. E. I. Langlois and G. A. Gresham, The aging of bruises: a review
and study of the colour changes with time,Forensic Sci. Int. 50, 227
238 (1991).
Journal of Biomedical Optics 030901-9 March 2015 Vol. 20(3)
Wilson et al.: Review of short-wave infrared spectroscopy and imaging methods.. .
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/02/2015 Terms of Use: http://spiedl.org/terms
44. S. Maguire et al., Can you age bruises accurately in children? A
systematic review,Arch. Dis. Child. 90, 187189 (2005).
45. L. Wang, S. L. Jacques, and L. Zheng, Monte Carlo modeling of pho-
ton transport in multi-layered tissues,Comput. Methods Programs
Biomed. 47, 131146 (1995).
46. R. M. Schols et al., Automated spectroscopic tissue classification in
colorectal surgery,Surg. Innovation 23 (2015).
47. S. Waxman, F. Ishibashi, and J. D. Caplan, Rationale and use of
near-infrared spectroscopy for detection of lipid-rich and vulnerable
plaques,J. Nucl. Cardiol. 14(5), 719728 (2007).
48. C. M. Gardner et al., Detection of lipid core coronary plaques in
autopsy specimens with a novel catheter-based near-infrared spectros-
copy system,JACC Cardiovasc. Imaging 1(5), 638648 (2008).
49. J. A. Goldstein et al., Detection of lipid-core plaques by intracoronary
near-infrared spectroscopy identifies high risk of periprocedural myocar-
dial infarction,Circ. Cardiovasc. Interv. 4(5), 429437 (2011).
50. Infraredx(R), http://www.infraredx.com/ (9 March 2015).
51. P. Taroni et al., Clinical trial of time-resolved scanning optical mam-
mography at 4 wavelengths between 683 and 975 nm,J. Biomed. Opt.
9(3), 464473 (2004).
52. A. Cerussi et al., In vivo absorption, scattering, and physiologic
properties of 58 malignant breast tumors determined by broadband
diffuse optical spectroscopy,J. Biomed. Opt. 11(4), 044005
(2006).
53. E. D. Jansen et al., Temperature dependence of the absorption coeffi-
cient of water for midinfrared laser radiation,Lasers Surg. Med. 14(3),
258268 (1994).
54. N. Kakuta et al., Temperature imaging of sub-millimeter thick water
using a near infrared camera,Int. J. Heat Mass Transfer 52(1920),
42214228 (2009).
55. T. L. Troy, D. L. Page, and E. M. Sevick-Muraca, Optical properties of
normal and diseased breast tissues: prognosis for optical mammogra-
phy,J. Biomed. Opt. 1(3), 342355 (1996).
56. J. Laufer et al., Effect of temperature on the optical properties of
ex vivo human dermis and subdermis,Phys. Med. Biol. 43(9),
24792489 (1998).
Biographies of the authors are not available.
Journal of Biomedical Optics 030901-10 March 2015 Vol. 20(3)
Wilson et al.: Review of short-wave infrared spectroscopy and imaging methods.. .
Downloaded From: http://biomedicaloptics.spiedigitallibrary.org/ on 04/02/2015 Terms of Use: http://spiedl.org/terms
... Analysis of the major peaks in regression coefficients identified significant wavelengths (1140, 1220, and 1400 nm) ( Figure 3). samples [46]. The spectral peaks observed at 1100, 1200, and 1400 nm align with known molecular changes in food tissues, making these wavelengths promising indicators for quality classification. ...
... Significant differences between RT and WT samples were observed in the physicochemical properties and TPA results, with additional distinctions evident in HSI spectral data. PLS-DA of the HSI data demonstrated high classification accuracy (maximum Rc 2 = 0.9547) from day 1 to day 3 of storage, identifying the wavelengths of 1100, 1200, and 1400 The 1100 nm wavelength in SWIR corresponds to the second overtone of C-H stretching vibrations, which are essential for detecting organic molecules in biological tissues [46]. The 1200 nm wavelength, similar to the second overtone of C-H stretching, is valuable for assessing the presence of lipids and fats in food products, providing insights into fat content and composition [47]. ...
... Wilson et al. [46] noted that the SWIR range is particularly advantageous for noninvasive tissue analysis, since its deeper penetration compared to that of visible light allows for more thorough examination of tissue composition changes. In this study, the 1100 nm wavelength correlated with variations in lipid content within the tissue, whereas the changes at 1200 nm appear to reflect muscle oxygenation status, potentially indicating variations in tissue perfusion [49]. ...
Article
Full-text available
Seafood quality preservation remains a critical focus in the food industry, particularly as the freeze–thaw process significantly impacts the freshness and safety of aquatic products. This study investigated quality changes in frozen mackerel subjected to two thawing methods, room temperature (RT) and running water (WT), and assessed the potential of hyperspectral imaging (HSI) for classifying these methods. After thawing, mackerel samples were stored at 5 °C for 21 days, with physicochemical, textural, and spectroscopic analyses tracking quality changes and supporting the development of a spectroscopic classification model. Compared with the WT method, the RT method delayed changes in key quality indicators, including pH, total volatile basic nitrogen (TVB-N), and total viable count (TVC), by 1–2 days, suggesting it may better preserve initial quality. Texture profile analysis showed similar trends, supporting the benefit of RT in maintaining quality. A major focus was on using HSI to assess quality and classify thawing methods. HSI achieved high classification accuracy (Rc2 = 0.9547) in distinguishing thawing methods up to three days post-thaw, with 1100, 1200, and 1400 nm wavelengths identified as key spectral markers. The HIS’s ability to detect differences between thawing methods, even when conventional analyses showed minimal variation, highlights its potential as a powerful tool for quality assessment and process control in the seafood industry, enabling detection of subtle quality changes that traditional methods may miss.
... The optical parameters of the seven skin layers across the 1000-1700 nm wavelength range are shown in Figure 2. To explore the selective detection of skin depth using multi-SDSs, we proposed circular photosensitive area detectors with a radius of 0.1 mm, placed on the skin surface at specific separations from the light source. Specifically, we selected wavelengths of 1000 nm and 1500 nm, where hemoglobin and water exhibit strong absorption [37]. The proportion of the mean optical path in each skin layer relative to the total mean optical path is shown in Figure 3. ...
Article
Full-text available
Optical sensors serve as pivotal components in the realm of non-invasive blood glucose measurement (NBGM) devices, where their efficacy directly influences the detection of weak glucose signals. This research introduces three fundamental design key points tailored for diffuse reflectance optical sensors employed for NBGM: depth resolution, detection signal-to-noise ratio, and human–sensor interface coupling. Guided by these design key points, we presented feasible design proposals for near-infrared diffuse reflectance sensors operating in the range of 1000–1700 nm. As an example, a sensor composed of five-ring detectors with a ring-shaped mask were made and tested on human skin. The innovative sensor developed herein holds promising potential for NBGM.
... In the absence of optical properties at the infrared wavelength ($1,875 nm) for in vivo brain tissue, we adapted the light absorption and scattering properties of skin measured by experiments (absorption, 1.97; scattering, 12.5). 16,37,38 The main irradiance distribution (90%) modeled from a 200-mm fiber apposed to the cortical surface and delivering 0.5 J/cm 2 (Figure 3A, middle) was within a diameter of 300 mm at all depths ( Figure 3A, see light irradiance). ...
... Broadband fiber laser sources are widely used in fluorescence imaging [1], agricultural [2], optical power limiter [3], and biological tissue characterization [4]. There are two commonly used methods for obtaining broadband sources in fiber laser systems. ...
Article
Full-text available
The Mamyshev oscillator (MO) is a nonlinear system that achieves mode-locking through the spectral broadening caused by self-phase modulation and the spectral filtering effect. However, excess nonlinear accumulation in MO will lead to pulse splitting, which limits the generation of high-energy pulses and broadband spectra. Therefore, how to design an all-fiber MO so that it can not only achieve mode-locking, but also tolerate strong nonlinear to generate broad-spectrum optical pulses is important, but has not been reported. Here, combining the perfect saturable absorption effect provided by the MO and the self-similar pulse (SSP) that can tolerate strong nonlinearity without wave breaking, we construct a theoretical model to investigate the ultrabroadband SSP that breaks through the gain bandwidth limit in a 2-µm MO. Through systematically analyzing the effects of fiber length, filter bandwidth, filter separation, and gain saturation energy on the spectral width of the output pulse, using the optimal parameters, we obtained a 3-dB ultra-broad spectrum of 290 nm, which is the widest bandwidth achieved for single pulse obtained in thulium-doped fiber lasers. Benefiting from the linear chirp characteristics of the SSPs, a dechirped pulse of 37.54 fs (\sim 6 cycles) has been achieved. This work provides theoretical guidance for experimentally obtaining ultrabroadband, few-cycle optical pulses in fiber lasers.
Chapter
Nanotechnology is an emerging technology with specialized nanomaterials and devices proficient for agriculture harvest. Nanomaterials can play important role in agriculture for improving the crop productivity in sustainable manner. The nanomaterials are produced by green synthesis and artificial synthesis. The benefits of green synthesis lower energy use, lower costs, and the utilization of renewable ingredients, whereas, use of artificially synthesized nanomaterials in agriculture has increased due to their ability to increase crop yield, improve soil health, and minimize dependency on traditional chemical inputs. Nanotechnology provide micronutrient bioactive substances enhanced nutrients absorption, disease control, pest and weed management, prevention of nutrient loss improve the agricultural yield. The core, types, composition, size, shape, and chemistry of nanomaterials play an important role in various agriculture tasks such as nanofertilizers, nanopesticides, nanoherbicides, nanoinsecticides, and nanosensors. In this chapter, we discuss the selection and application of suitable nanomaterials to achieve sustainable agriculture and remedies of nanotoxicity. In agriculture, nanotechnology can bring revolutionary changes in harvest, weed and pest management, soil-plant interaction, and food and science technology.
Article
Full-text available
Background. In colorectal surgery, detecting ureters and mesenteric arteries is of utmost importance to prevent iatrogenic injury and to facilitate intraoperative decision making. A tool enabling ureter- and artery-specific image enhancement within (and possibly through) surrounding adipose tissue would facilitate this need, especially during laparoscopy. To evaluate the potential of hyperspectral imaging in colorectal surgery, we explored spectral tissue signatures using single-spot diffuse reflectance spectroscopy (DRS). As hyperspectral cameras with silicon (Si) and indium gallium arsenide (InGaAs) sensor chips are becoming available, we investigated spectral distinctive features for both sensor ranges. Methods. In vivo wide-band (wavelength range 350-1830 nm) DRS was performed during open colorectal surgery. From the recorded spectra, 36 features were extracted at predefined wavelengths: 18 gradients and 18 amplitude differences. For classification of respectively ureter and artery in relation to surrounding adipose tissue, the best distinctive feature was selected using binary logistic regression for Si- and InGaAs-sensor spectral ranges separately. Classification performance was evaluated by leave-one-out cross-validation. Results. In 10 consecutive patients, 253 spectra were recorded on 53 tissue sites (including colon, adipose tissue, muscle, artery, vein, ureter). Classification of ureter versus adipose tissue revealed accuracy of 100% for both Si range and InGaAs range. Classification of artery versus surrounding adipose tissue revealed accuracies of 95% (Si) and 89% (InGaAs). Conclusions. Intraoperative DRS showed that Si and InGaAs sensors are equally suited for automated classification of ureter versus surrounding adipose tissue. Si sensors seem better suited for classifying artery versus mesenteric adipose tissue. Progress toward hyperspectral imaging within this field is promising.
Article
Full-text available
Extending the wavelength range of spatial frequency domain imaging (SFDI) into the short-wave infrared (SWIR) has the potential to provide enhanced sensitivity to chromophores such as water and lipids that have prominent absorption features in the SWIR region. Here, we present, for the first time, a method combining SFDI with unstructured (zero spatial frequency) illumination to extract tissue absorption and scattering properties over a wavelength range (850 to 1800 nm) largely unexplored by previous tissue optics techniques. To obtain images over this wavelength range, we employ a SWIR camera in conjunction with an SFDI system. We use SFDI to obtain in vivo tissue reduced scattering coefficients at the wavelengths from 850 to 1050 nm, and then use unstructured wide-field illumination and an extrapolated power-law fit to this scattering spectrum to extract the absorption spectrum from 850 to 1800 nm. Our proof-of-principle experiment in a rat burn model illustrates that the combination of multispectral SWIR imaging, SFDI, and unstructured illumination can characterize in vivo changes in skin optical properties over a greatly expanded wavelength range. In the rat burn experiment, these changes (relative to normal, unburned skin) included increased absorption and increased scattering amplitude and slope, consistent with changes that we previously reported in the near-infrared using SFDI. (C) The Authors. Published by SPIE under a Creative Commons Attribution 3.0 Unported License.
Article
Full-text available
Optical frequency domain imaging (OFDI) can identify key components related to plaque vulnerability but can suffer from artifacts that could prevent accurate identification of lipid rich regions. In this paper, we present a model of depth resolved spectral analysis of OFDI data for improved detection of lipid. A quadratic Discriminant analysis model was developed based on phantom compositions known chemical mixtures and applied to a tissue phantom of a lipid-rich plaque. We demonstrate that a combined spectral and attenuation model can be used to predict the presence of lipid in OFDI images.
Article
Time-domain diffuse optical spectroscopy is being applied with increasing success to study highly scattering media, mainly in the fields of non-invasive medical diagnostics and quality assessment of food and pharmaceutical products. The region beyond 1100nm is still largely unexplored by time-domain techniques, probably due to the difficult combination of tunable pulsed sources and suitable single-photon detectors. We extend the spectral range up to 1700 nm thanks to a pulsed supercontinuum laser and a time-gated InGaAs/InP single-photon avalanche diode, with potential applications in medical diagnostics and in the study of scattering materials. A first application on lipids is shown.
Article
Background— A method is needed to identify nonstenotic, lipid-rich coronary plaques that are likely to cause acute coronary events. Near-infrared (NIR) spectroscopy can provide information on the chemical composition of tissue. We tested the hypothesis that NIR spectroscopy can identify plaque composition and features associated with plaque vulnerability in human aortic atherosclerotic plaques obtained at the time of autopsy. Methods and Results— A total of 199 samples from 5 human aortic specimens were analyzed by NIR spectroscopy. Features of plaque vulnerability were defined by histology as presence of lipid pool, thin fibrous cap (<65 μm by ocular micrometry), and inflammatory cell infiltration. An InfraAlyzer 500 spectrophotometer was used. Spectral absorbance values were obtained as log (1/R) data from 1100 to 2200 nm at 10-nm intervals. Principal component regression was used for analysis. An algorithm was constructed with 50% of the samples used as a reference set; blinded predictions of plaque composition were then performed on the remaining samples. NIR spectroscopy sensitivity and specificity for histological features of plaque vulnerability were 90% and 93% for lipid pool, 77% and 93% for thin cap, and 84% and 89% for inflammatory cells, respectively. Conclusions— NIR spectroscopy can identify plaque composition and features associated with plaque vulnerability in postmortem human aortic specimens. These results support efforts to develop an NIR spectroscopy catheter system to detect vulnerable coronary plaques in living patients.
Article
Non-invasive and non-destructive Near Infrared Reflectance Spectroscopy (NIRS) analysis of pork quality parameters is challenging due to the presence of a structured skin layer. The optical characterization (absorption coefficient, scattering coefficient and anisotropy factor) of porcine skin tissue can help in a better understanding of the light-tissue interaction, more efficient instrumental sensor designs and data modeling. For this purpose, double integrating spheres combined with an unscattered transmittance measurement system working in the range 1150-2250 nm (interval step of 10 nm) were used for optical characterization of pig epidermis and dermis. The estimated pig skin absorption coefficient spectra were dominated by water, while scattering coefficient spectra showed a steady decreasing tendency with increasing wavelength, except at strong absorption bands. The anisotropy factor spectra showed a stable value for each layer, although also some signatures were observed at strong absorption bands. Industrial relevance: A novel double integrating sphere system combined with an unscattered transmittance detector based both on a supercontinuum laser is used for optical characterization (absorption, scattering and angular scattering distribution) of pig skin tissues (epidermis and dermis) in the 1150-2250 nm wavelength range. The estimated data reported are relevant to understand individual effects of absorption, scattering and angular scattering distribution of NIR light through the tissue. Moreover, it can be used for simulating non-destructive NIRS measurement what can have an impact in the advance of NIRS technology toward in vivo/intact analysis.
Article
Optical diagnostics of bruised skin might provide important information for characterization and age determination of such injuries. Hyperspectral imaging is one of the optical techniques that have been employed for bruise characterization. This technique combines high spatial and spectral resolution and makes it possible to study both chromophore signatures and -distributions in an injury. Imaging and spectroscopy in the visible spectral range have resulted in increased knowledge about skin bruises. So far the SWIR region has not been explored for this application. The main objective of the current study was to characterize bruises in the SWIR wavelength range. Hyperspectral images in the SWIR (950-2500nm ) and VNIR (400-850nm) spectral range were collected from 3 adult volunteers with bruises of known age. Data were collected over a period of 8 days. The data were analyzed using spectroscopic techniques and statistical image analysis. Preliminary results from the pilot study indicate that SWIR hyperspectral imaging might be an important supplement to imaging in the visible part of the spectrum. The technique emphasizes local edema and gives a possibility to visualize features that cannot easily be seen in the visible part of the spectrum.
Article
This study aimed to evaluate the potential of NIR spectroscopy as a non-destructive tool for predicting fatty acid (FA) composition of different sturgeon’s caviars; Asetra, Beluga and Sevruga. Partial least squares regression (PLSR) models were built under cross-validation and tested with independent data sets, yielding determination of coefficients (R2) between 0.81 and 0.97 with standard errors of 0.08 to 2.90. Selected models for the estimation of major constituents (C14:0, C14:1, C16:0, C16:1, C18:0, C18:1, C18:3, C20:5 and C22:6) and saturated (SFA), monounsaturated (MUFA) and polyunsaturated fatty acid (PUFA) groups were also determined in caviars.