Miniature endoscope for simultaneous optical
coherence tomography and laser-induced
fluorescence measurement
Alexandre R. Tumlinson, Lida P. Hariri, Urs Utzinger, and Jennifer K. Barton
We have designed a multimodality system that combines optical coherence tomography 共OCT兲 and
laser-induced fluorescence 共LIF兲 in a 2.0-mm-diameter endoscopic package. OCT provides ⬃18-␮m
resolution cross-sectional structural information over a 6-mm field. LIF spectra are collected sequentially at submillimeter resolution across the same field and provide histochemical information about the
tissue. We present the use of a rod prism to reduce the asymmetry in the OCT beam caused by a
cylindrical window. The endoscope has been applied to investigate mouse colon cancer in vivo. © 2004
Optical Society of America
OCIS codes: 120.3890, 120.5800, 170.2150, 170.2680, 170.4500, 170.6280.
1. Introduction
Colorectal cancer is the third most common type of
cancer in the United States and is expected to cause
10% of cancer deaths in 2003.1 Advanced tools to
monitor the early stages of this disease in humans
and animal research models are needed to effectively
combat this killer. Multiple researchers have demonstrated the use of laser-induced fluorescence 共LIF兲
to differentiate normal tissue from cancer and precancer tissue in the human colon using only endogenous fluorophores. Richards-Kortum et al.2 found
excitation wavelengths at 330, 370, and 430 ⫾ 10 nm
to be the most useful for discriminating normal from
adenotomatous tissues in vitro. At an excitation of
370 nm, fluorescence emission intensities at 404, 480,
and 680 nm were found to be most informative.2
Schomacker et al.3 noted significantly different emission spectra in vivo and in vitro due to the exponential decay of the reduced form of nicotinamide
adenine dinucleotide after resection, yet achieved
good sensitivity and specificity at 337-nm excitation
All the authors are with the University of Arizona, 1230 East
Speedway Boulevard, Tucson, Arizona 85721-0104. A. R. Tumlinson and J. K. Barton 共barton@u.arizona.edu兲 are with the Optical Sciences Center. L. P. Hariri, U. Utzinger, and J. K. Barton
are with the Division of Biomedical Engineering.
Received 16 April 2003; revised manuscript received 2 September 2003.
0003-6935兾04兾010113-09$15.00兾0
© 2004 Optical Society of America
for both cases. Computer simulation has indicated
that the fluorescence changes observed in adenoma
are a result of alteration in both architecture and
biochemistry.4 To our knowledge, no research has
been published that examines endogenous fluorophores in small animal colon cancer models.
High-resolution cross-sectional images of the human colon have been collected by optical coherence
tomography 共OCT兲. Kobayashi et al.5 reported visualizing crypts, the muscularis mucosae, fibrous tissue, blood vessels, and lymphatic nodules in the
submucosa. Pitris et al.6 demonstrated the OCT’s
ability to visualize the structural differences between
the organization of normal colon and the disorganization in tissues from ulcerative colitis and adenocarcinoma.6 Sivak et al.7 found that, with the use of an
endoscopic OCT, features such as colonic crypts were
best seen when the probe was held a small distance
away from the tissue surface. Deeper structures
such as muscularis mucosa and submucosa were better visualized when the probe was held against the
gut wall.7
The use of OCT and LIF together to identify cancer
is just emerging as a possible diagnostic technique.
To detect oral cancer, McNichols et al.8 developed a
large-diameter 共⬃1.5-cm兲 rigid endoscopic probe in
which autofluorescence imaging was used as a
method of rapidly identifying suspicious regions to
subsequently inspect with OCT for more definitive
morphological information.8 Kuranov et al.9 used
separate OCT and LIF 共the fluorescence sensitizer
5-aminolevulinic acid was administered systemical1 January 2004 兾 Vol. 43, No. 1 兾 APPLIED OPTICS
113
ly兲 probes to examine the borderline regions of cervical cancer. They found that the combination of
these two modalities produced fewer false positive
results: Increased 5-aminolevulinic acid fluorescence due to inflammatory reactions was clearly differentiated from cancer by OCT; conversely, a lack of
structure in a mature scar indicated by OCT was
clarified by LIF.9,10 In vitro studies of mouse colon
adenomas performed in our laboratory with a bulkoptic combined OCT–LIF system indicated results
consistent with research on human tissue. At
325-nm excitation, we observed a dramatic decrease
in the ratio of 385– 480-nm fluorescence intensity
over the adenoma, whereas OCT indicated a loss of
organization in the same area.11
Several OCT endoscope geometries have been described in the literature. The first type achieves a
radial scan by transmitting a rotation through an
inner cable to the optics at the distal end. A rotary
optical coupler allows the fiber to spin and maintain
alignment in a focused beam.12 The coupling
scheme is simplified in a second type of endoscope,
which pushes and pulls an inner cable to transmit a
linear translation to the distal optics.13 In a recent
type of endoscope, a microelectromechanical system
mirror is used to direct an end scan.14 Flexible fluorescence spectroscopy probes are typically a simple
fiber or fiber bundle held near or in contact with the
tissue. A wide variety of illumination geometries
have been used15 that results in different collection
efficiencies16 and tissue sampling depths.17,18
The C57BL兾6J-ApcMin mouse strain 共Min mouse兲
has a genetic mutation of the mApc gene causing it to
spontaneously develop tumors in the gastrointestinal
tract. This phenotype makes the Min mouse a suitable model for studying human colorectal cancer19
and has been widely used to study the effects of genetics,20 exercise,21 diet,22 and therapeutic drugs23 on
this disease. Colonoscopies have been performed on
this type of mouse with a 2.1-mm-diameter pediatric
cystoscope with an average insertion depth of 3 cm
without unusual complications.24
We built a dual-purpose endoscopic instrument
that produces both high-resolution OCT images and
fluorescence spectra. In this paper we describe our
instrument for obtaining simultaneous LIF spectra
and OCT images and present preliminary results obtained from an in vivo mouse colon.
selection a primary concern. LIF systems tend to be
sensitive to changes in illumination geometry; therefore a diffuse beam is used at a distance that is constrained by the packaging of the endoscope.25
A diagram of the experimental setup is shown in
Fig. 1. The OCT and LIF portions of the system
were optically distinct except for the final fold mirror
and exit window. Analog processing electronics for
the subsystems were distinct but controlled by a central computer.
2. Materials and Methods
B.
A.
Dual-Modality Instrument
The technical challenge in combining OCT and LIF
endoscopically is to meet the drastically different requirements of each system in the same minute package. OCT requires a focused near-IR beam for deep
共⬃2-mm兲 tissue penetration. LIF usually excites
the tissue with a monochromatic beam of relatively
short- 共UV– green兲 wavelength light with shallow tissue penetration 共⬃200 ␮m for 325 nm兲. The design
of a LIF system must be optimized at each element to
minimize autofluorescence, which makes materials
114
APPLIED OPTICS 兾 Vol. 43, No. 1 兾 1 January 2004
Fig. 1. Overview of the combined OCT–LIF endoscope shows two
systems that meet at the endoscope tip and at the data-acquisition
computer. The OCT system consists of a superluminescent diode
共SLD兲 source, coupled through a fiber-optic beam splitter 共50兾50兲 to
the sample arm, the reference arm 关polarization controller 共PC兲,
fiber-optic collimator 共FC兲, and reference mirror 共RM兲兴, and the
detector 共DET兲. The LIF system consists of a He–Cd excitation
laser 共EL兲 that is selectable for 325- or 442-nm output radiation
and a neutral-density filter 共ND兲 to reduce the intensity before the
light is coupled into the excitation channel by a focusing lens 共FL兲.
Emission light returning on the collection fibers passes a collimating lens 共CL兲 before the excitation light is removed by a long-pass
filter 共LP兲, and a focusing lens 共FL兲 images the fibers onto the
entrance slit of the spectrometer 共S兲. The OCT sample arm, LIF
excitation, and LIF collection fibers are bundled together into the
inner lumen of the endoscope body. This inner lumen is pushed
and pulled relative to the outer lumen by the motion controller
共MC兲 to cause a lateral scan at the endoscope tip optics 共ETO兲.
The computer 共COM兲 interfaces with the OCT system through the
data-acquisition board 共DAQ兲, which controls the reference mirror
共RM兲, receives image data from the lock-in amplifier 共LIA兲, and
controls the lateral scan of the tip optics. The computer interfaces
with the LIF system through the general-purpose interface bus
共GPIB兲 interface to the spectrometer controller 共SC兲, which controls the setup of the spectrometer 共S兲 and gathers data from the
CCD.
Optical Coherence Tomography Subsystem
The OCT subsystem was described in detail previously.26 A superluminescent diode source was coupled into the source arm of a fiber Michelson
interferometer. A 1300-nm center wavelength and
49-nm bandwidth yielded a coherence length of approximately 16 ␮m in air 共11 ␮m in tissue, assuming
an average of n ⫽ 1.4兲. The reference arm consisted
of a simple, slow galvanometer-mounted retroreflector providing 2 mm of path-length modulation at 14
a-scans兾s 共image columns per second兲. The sample
arm optics are described in Subsection 2.D. Reflected light from the sample and reference arms was
recombined at the fiber-optic beam splitter before being measured by a photodetector with a built-in bandpass filter. The interference between the two beams
could be observed when the optical paths matched
within a coherence length of the source. The resulting signal, a 98-kHz heterodyne beat due to the OCT
source center wavelength and the velocity of the reference mirror, was demodulated with a lock-in amplifier and digitized with a data-acquisition board.
C. Laser-Induced Fluorescence Spectroscopy
Subsystem
The LIF subsystem used a He–Cd laser 共Kimmon
Electric, Englewood, Colorado兲 as an excitation
source, operated at 325 nm 共15 mW兲 or 442 nm 共72
mW兲. A narrowband filter on the laser selected the
output wavelength. A metalized neutral-density filter reduced the laser power before the light was coupled into an aluminum-jacketed, all-silica optical
fiber, with a 200-␮m core and a 0.22 numerical aperture 共Fiberguide, Stirling, New Jersey兲 using a
150-mm focal-length fused-silica singlet. Light
emitted from the sample was returned for measurement by two fibers similar to the one just described.
The prespectrometer optics consisted of a 35-mm
focal-length fused-silica singlet to collimate the fiber
output, a filter to remove the excitation light 共325 nm:
dielectric-coated long pass, 333AELP, Omega Optical, Brattleboro, Vermont; 442 nm: color glass long
pass, GG455, Schott Glas, Mainz, Germany兲, and a
75-mm focal-length BK-7 singlet to focus the light
onto the spectrometer entrance slit. This arrangement removed the reflected laser line from the spectra and matched the numerical aperture of the fiber
and spectrometer. Spectra were measured with a
modified Czerny–Turner grating spectrometer 共Triax
180, Jobin-Yvon, Edison, New Jersey兲 with a cooled
CCD 共Spectrum One mini-thermoelectric air-cooled
CCD, Jobin-Yvon兲. The 300-lines兾mm grating was
centered near 500 nm, giving a range of 300 –750 nm
without scanning the grating. The entrance slit was
opened to 500 ␮m, allowing nearly all the light from
the return fibers into the spectrometer and limiting
the spectrometer resolution to 5 nm FWHM. A universal serial bus to IEEE488 interface 共National Instruments,
Austin,
Texas兲
controlled
the
spectrometer setup and the transfer of measured values from the CCD controller to the computer.
D.
Endoscope Tip Optics
The sample arm optics were designed so that the
OCT and LIF excitation beams were nearly coaligned
on the sample as can be seen in Fig. 2共a兲. Code V
optical modeling software 共Optical Research Associates, Pasadena, California兲 was used to characterize
beam propagation of both optical paths through the
tip optics. Gaussian beam trace options were used
to determine the correct length of the gradient-index
共GRIN兲 lens and characterize the waist asymmetry
introduced by the cylindrical output window. To explore this asymmetry more generally, an on-axis
asphere was modeled to create an aberration-free 15-
Fig. 2. 共a兲 Code V model of the endoscope tip shows the focused IR
for OCT as well as the diffuse UV illumination. IR illumination
for OCT is launched from a single-mode fiber 共O兲, focused by a
GRIN lens 共G兲, reflected sideward by a rod prism 共P兲, through a test
tube window 共W兲, and into the tissue 共T兲. UV illumination for LIF
is launched from a multimode fiber 共E兲 and diffuses to a wide spot
before hitting the tissue. 共b兲 A solid model of the endoscope tip
illustrates how the LIF excitation fiber 共E兲 and collection fibers 共C兲
sit above the OCT channel 关the fiber 共O兲, stripped of its jacket,
passes through the ferrule 共F兲 to contact the GRIN lens 共G兲兴. Both
systems are cemented to a rod prism 共P兲 that reflects the light
through the side of a thin-walled silica tube 共W兲. 共c兲 An image of
the endoscope with the optics retracted superimposed upon an
image with the optics translated to the tip.
␮m-diameter spot in tissue, and a variety of prisms
and windows were inserted to explore the effects of
these elements in configurations that may be interesting outside of this specific application. Pointspread function analysis was used to determine the
imaged spot quality, summarized in the Strehl ratio.
Simple ray tracing was used to identify areas where
total internal reflection 共TIR兲 might block the beam
propagation. Beam footprint analysis was used to
visualize the amount of overlap between the emission
and the collection areas on the tissue.
1 January 2004 兾 Vol. 43, No. 1 兾 APPLIED OPTICS
115
Figure 2共b兲 shows the mechanical arrangement of
the endoscope tip optics. A silica ferrule 共InnovaQuartz, Phoenix, Arizona兲 centered the OCT fiber
共Corning SMF-28, Thorlabs兲 on the GRIN lens 共SLW
1.0 050 130, NSG America, Sommerset, New Jersey兲,
cut to 3.48 mm. This GRIN lens was cemented to a
1.8-mm-diameter, 2.2-mm-long rod prism with an
aluminized 45-deg face 共InnovaQuartz兲 that acted as
a fold mirror to direct the beam out the side of a
fused-silica test tube with an outer diameter of 2.0
mm and a wall thickness of 100 ␮m 共InnovaQuartz兲.
The excitation fiber was situated between the two
emission collection fibers. These three fibers sat directly above the ferule and GRIN lens of the OCT
path and were cemented coplanar with the GRIN lens
to the rod prism. All optical surfaces were joined
with low-fluorescence UV curing epoxy 共Norland 63,
Norland Products, Cranbury, New Jersey兲. The cylindrical surface of the GRIN lens near the fluorescence fibers was blackened with a permanent marker
to reduce the risk of autofluorescence from the GRIN
material.
E.
Mechanical Translation and Endoscope Body
We achieved a linear scan by translating the endoscope tip optics through the length of the test tube
window as demonstrated in Fig. 2共c兲. A voltage
supplied by an analog output board 共National Instruments兲 provided the signal to the interface
board and linear galvanometer 共GSI Lumonics, Wilmington, Massachusetts兲, which created the driving
force for the scan. The translation generated at
the galvo was transmitted to the tip by a
polytetrafluoroethylene-coated polyimide tube 共Microlumen, Tampa, Florida兲 containing the optical
fibers. This assembly slid through a 135-cm-long
outer polyimide tube 共Microlumen兲 that was cemented 共Norland 63, Norland Products兲 to the test
tube window. We maintained the four fibers parallel inside the inner-cable lumen by carefully applying 2-cm sections of thin-walled polyester shrink
tube 共Advanced Polymers, Salem, New Hampshire兲
at approximately 3-cm intervals. The length of a
lateral scan was limited to approximately 6 mm by
the test tube window. Mechanical and software
limits were applied to prevent the endoscope tip
optics from contacting the ends of the tube.
F.
Software, Data Acquisition, and Correction
LabVIEW software 共National Instruments兲 was used
to synchronize reference arm movement, sampling
optics lateral translation, spectrometer control, OCT
data acquisition, and LIF spectra measurement.
The system acquired data for OCT and one laser
excitation wavelength simultaneously. The spectrometer was set to take data with an exposure time
of 1 s and a minimum delay 共⬃500 ms due to data
transfer and computation time兲 between exposures.
The CCD image was binned in the slit direction to
yield a single spectrum. The spectral acquisition
system was calibrated for wavelength with a mercury
116
APPLIED OPTICS 兾 Vol. 43, No. 1 兾 1 January 2004
Table 1. Concentrations of Reference Dyes Used
Dye
Solvent
Desired
Volume 共␮l兲 of 1-mM
Concentration Stock Solution into
共␮M兲
50-ml Solvent
Quinine
0.5-M H2SO4
sulphate
Fluorescein 0.1-M NaOH
Sulphorho- Ethanol
damine
2.0
100
0.2
0.8
10
40
line source 共Pen-Ray Lamp, UVP, Upland, California兲.
MATLAB 共The MathWorks, Natick, Massachusetts兲
software was used off line to apply spectral correction
factors and visualize the collected data. These corrections remove known environmental effects from
the signal and correct for nonuniform spectral response due to transmission of optical components and
detector response. CCD dark current and analog-todigital converter offset were subtracted from all spectra. A separate room-light background measured
with the laser off and the endoscope inserted in the
mouse was subtracted from in vivo spectra. An
autofluorescence background was estimated by insertion of the endoscope into a bath of distilled water
with the laser on and was subtracted from in vivo
data. Spectral sensitivity, measured with a National Institute of Standards and Technologytraceable tungsten light source 共LS-1-CAL, Ocean
Optics, Dunedin, Florida兲, was divided through all
measured spectra. Spectral data were acquired every 0.4 nm compared with a system resolution of 5 nm
FWHM. Therefore all spectra were filtered to match
the system resolution with a Savitzky–Golay filter.
Spectral readings were coregistered with OCT readings, accounting for the offset between the beams in
the tip optics and the scan distance covered during
spectra acquisition. The fluorescence intensity was
sampled at three wavelengths near the peaks in typical tissue spectra and displayed as a bar chart under
corresponding sections of the OCT image. We calculated the LIF signal-to-noise ratio using spectra
from live tissue. The pixel-to-pixel noise is assumed
to be similar to the noise from spectra to spectra,
which allows us to compare the difference between
the unfiltered signal and the signal after the
Savitzky–Golay filter was applied to the signal
strength in the same spectral region.
G.
Reference Dye Spectra
We determined the LIF performance of our device by
comparing reference dye 共quinine, fluorescein, and
rhodamine兲 共Molecular Probes, Eugene, Oregon兲 spectra collected with a commercial spectrofluorometer
共Fluorolog 3-32, Jobin-Yvon Horiba兲 with those collected from the same dyes with our device. Reference
dye concentrations were determined 共see Table 1兲 such
that their fluorescent intensity when excited with 325and 442-nm light would be similar to that of human
skin when exposed to the same excitation. Diluted
Table 2. Various Configurations Modeled with a Code V Gaussian Beam Tracea
Waist Radius 共␮m兲
Material
ODb 共mm兲
IDc 共mm兲
Material
Geometry 共mm兲
X 共␮m兲
Y 共␮m兲
X:Y
Axial Separation
Depth x and
Depth y 共mm兲
Fused silica
2.0
1.8
Fused silica
3.1
1.1
Polyethylene
1.25
1.1
BK-7
Silica
MgF2
BK-7
Silica
MgF2
BK-7
Rectangular 1.0 共edge兲
Cylindrical 1.7 共diameter兲
Cylindrical 1.7
Rectangular 0.5
Cylindrical 1.0
Cylindrical 1.0
Rectangular 0.5
8.2
7.3
7.5
21.8
6.7
7.6
9.7
7.7
7.7
7.7
7.7
7.7
7.7
7.7
1.06
0.95
0.97
2.83
0.87
0.99
1.26
0.04
⫺0.03
⫺0.02
2.63
⫺0.30
⫺0.08
0.09
0.6
Silica
MgF2
BK-7
Cylindrical 1.0
Cylindrical 1.0
Rectangular 0.3
7.1
7.4
12
7.7
7.7
7.7
0.92
0.96
1.56
⫺0.06
⫺0.03
0.13
Silica
MgF2
Cylindrical 0.5
Cylindrical 0.5
6.3
6.9
7.7
7.7
0.82
0.90
⫺0.14
⫺0.09
Window Description
Polyethylene
Prism Description
0.75
a
Results indicate that a cylindrical rod prism helps reduce beam asymmetry created by a cylindrical output window immersed in a
high-index media such as tissue.
b
OD, outside diameter.
c
ID, inside diameter.
dyes were placed in 1-cm path-length quartz cuvettes
共Starna Cells, Atascadero, California兲, and fluorescence was characterized in a right-angle configuration
with the commercial spectrofluorometer. Our probe
was laid alongside the cuvette, and a drop of distilled
water was placed between the endoscope and cuvette
windows for index matching. A cuvette filled with
distilled water was also measured to estimate the
autofluorescence component.
H.
Imaging Live Mice
Two normal C57BL兾6J 共weight range, 26.2–26.4 g兲
and six colon cancer model C57BL兾6J-ApcMin 共24.6 –
29.4 g兲 mice 共Jackson Laboratories, Bar Harbor,
Maine兲 were imaged by the combined OCT–LIF endoscope at 18 weeks of age. Room lights inside the
laser curtain were turned off; however, some background light from the surrounding room remained.
Mice were anesthetized with 2.5% Avertin 共Sigma
Chemical Co., St. Louis, Missouri兲, administered intraperitoneally. The endoscope was coated with lubricant 共K-Y Jelly, McNeil-PPC, Skillman, New
Jersey兲 and inserted into the anus with the mouse in
a dorsal supine position. We acquired a survey of
the colon by sequentially taking images at three
depths 共23, 28, and 33 mm with respect to the anus兲
and rotating the mouse with respect to the endoscope
eight times in 45° increments. All mice were housed
by University Animal Care in microisolators on a
12:12-h light– dark cycle with free access to water and
standard laboratory chow. Protocols were approved
by the University of Arizona Institutional Animal
Care and Use Committee.
3. Results
A.
Endoscope Tip Modeling
The goal of the OCT path was to place the beam waist
0.3 mm into the tissue with a focused beam diameter
equal to the coherence length of the source 共11 ␮m,
n ⫽ 1.4兲. The long working distance required by the
packaging and the market availability of GRIN
lenses resulted in a focus of 18 ␮m in diameter. The
calculated Strehl ratio of 0.98 indicated that the imaged spot was of very high quality. Table 2 shows
how much beam asymmetry results from several different arrangements of cylindrical output windows
and beam-folding prisms. Table 2 shows that use of
a rod prism instead of a rectangular prism is most
helpful when the window is thick or the diameter of
the endoscope is small. The second case in Table 2
represents our arrangement with a thin silica window and a cylindrical silica prism. In this case, we
expect a 5% waist diameter asymmetry and an axial
waist separation of 30 ␮m. In the last case of Table
2, we predict results achievable with currently available 0.5-mm-diameter GRIN lenses and a thin polyethylene tubing window 共note that polyethylene
tubing has unacceptable fluorescence characteristics
for use in our device but may be a good option for
OCT-only devices.兲 By replacing the typical rectangular prism with a MgF2 rod prism, the beam-waist
diameter asymmetry is reduced from 56% to 10%,
and the axial separation between the waists is reduced from 0.13 to 0.09 mm. Because the radius of
the rod prism is smaller than the inner radius of the
tube window, the rod prism tends to overcorrect the
asymmetry.
The LIF path was examined for TIR and beam
footprint at the tissue. Side-firing OCT endoscopes
have traditionally been able to use TIR on the reflecting surface of the beam-folding prism; however, the
reflecting face of the rod prism was aluminized because TIR would not support the wide divergence
originating directly from the LIF fibers. Blocking
the excitation beam and the field of view of the emission collection fibers with TIR was a risk that had to
1 January 2004 兾 Vol. 43, No. 1 兾 APPLIED OPTICS
117
Fig. 3. Beam footprint analysis shows the excited spot 共dark circles兲 compared with the field of view of one of the emission fibers
共lighter circles兲. Multiple circles indicate a beam originating from
different locations across the 200-␮m fiber face. Jagged edges on
the right of the image indicate where steep refraction angles have
distorted the field of view of the emission fiber.
be evaluated where the beam exited the rod prism
and might be more problematic for smaller rod
prisms. Figure 3 shows the diameter and overlap of
the excitation and collection areas on the tissue.
The excitation light dispersed to a spot approximately 1.25 mm in diameter, and the collection area
appears to overlap more than 75% of the illuminated
area for each fiber. The numerical aperture of the
emission light collection system was limited by the
fibers to 0.22.
B.
Reference Dye Measurement
Dye measurements show excellent correspondence
between the commercial spectrofluorometer and our
device. The curves of Fig. 4 are normalized to match
the intensity at the fluorescein emission peak. The
difference between the two measurements is primarily due to autofluorescence in the endoscope. The
majority of the signal returned from distilled water is
autofluorescence and at its peak is seen to be 8% as
strong as the weakest dye peak.
C.
Animal Imaging
In vitro insertion of the endoscope into a mouse colon
共⬃10 weeks old兲 indicated that the endoscope could be
used at the prescribed depth without gross tissue
damage. In vivo images of mouse colon demonstrate
LIF and OCT 共Fig. 5兲 functionality. Each 6-mm image and spectra pair takes 45 s to acquire. The OCT
images clearly differentiated several distinct layers of
the colon, as well as nearby connective tissue in the
abdominal cavity. The most prominent features in
the area determined to be the colon were a moder118
APPLIED OPTICS 兾 Vol. 43, No. 1 兾 1 January 2004
Fig. 4. Emission spectra from reference dye solutions: quinine
共Q兲 共325-excitation only兲, fluorescein 共F兲, and rhodamine 共R兲. Dotted curves indicate spectra measured by a commercial fluorometer
in a right-angle configuration. Solid curves indicate spectra from
the same dyes measured with our device in a front face configuration. Spectral intensities between the two devices are normalized
at the fluorescein peak. The sharp peak at 650 nm is due to
second-order excitation light in the commercial instrument.
ately scattering homogeneous region corresponding
to the mucosa, a bright line indicating the boundary
between the mucosa and submucosa, and a second
bright line identifying the boundary between the tunica media and the adventitia. The sawtooth borders sometimes visible in the loose tissue of the
adventitia are an artifact due to the repetitive motion
of breathing. Many of the images show horizontal
structure deep in the image that can be explained by
a slow rotation of the polarization state of the reflected signal, caused by birefringent tissue such as
stretched collagen fibers, which results in a periodic
loss of contrast.
Emission spectra at 325-nm excitation showed
peaks near 390, 450, and occasionally 680 nm.
Typical unfiltered signal-to-noise ratio in tissue was
250. Figure 6 shows an example spectra from colon tissue compared with expected2,3 component
low the maximum permissible exposure limits for
skin set in the American National Standards Institute Z136.1 section 8 during data acquisition.
4. Discussion
Fig. 5. OCT image and associated LIF emission spectra taken
with the endoscope on mouse colon tissue. Bar chart 共middle兲
shows intensity of selected emission wavelengths 共390, 450, 680
nm兲 at corresponding positions in the image above. The OCT
image is 6 mm long ⫻ 1.0 mm deep. Features visible in the image
include glass 共g兲 of the endoscope window, mucosal layer 共m兲, the
boundary between the mucosa and the submucosa 共sm兲, the boundary between tunica media and adventitia 共tm兲, adventitia 共a兲, and
a motion artifact due to breathing 共b兲.
spectra: collagen, the reduced form of nicotinamide adenosine dinucleotide, flavin adenine dinucleotide, the reduced form of nicotinamide adenine
dinucleotide, and 共oxy兲hemoglobin 共absorbing兲. A
typical OCT and spectral pair showed spatially corresponding changes in spectra for changing thicknesses and structures seen in OCT; however, no
attempt has yet been made to generalize diagnostically meaningful corresponding changes in each
modality for this data set. In the above-described
experiment, the radiation exposure levels were be-
Fig. 6. Sample tissue spectrum is compared with likely component spectra. NADH, the reduced form of nicotinamide adenine
dinucleotide; FAD, flavin adenine dinucleotide; Hb, hemoglobin.
The research presented here demonstrates the successful combination of OCT and LIF in a single endoscopic package. OCT and LIF performance are
comparable to probes built solely for each modality.
The dramatically different requirements for these
modalities resulted in a design that is essentially two
parallel optical systems. Because of autofluorescence concerns, the traditional materials used in OCT
endoscopes were not suitable for use in this device.
The GRIN lens was chosen to make the manufacture
of the endoscope tip11 practical; however, it will fluoresce when illuminated with UV light, so it remains
as a mild autofluorescence risk despite the fact that it
is held away from the direct LIF path. Methods of
further shielding the GRIN lens or replacing it with a
system of ball or drum lenses may reduce the
autofluorescence to negligible levels. A novel aspect
of our endoscope design is the use of a rod prism to
fold the beam into a side-firing configuration. In all
OCT endoscopes with cylindrical output windows, the
surface of the tube window and the surrounding tissue act as a negative cylindrical lens. The positive
cylinder exit surface of a rod prism can reduce asymmetry in the OCT beam when compared with a similar arrangement in which a rectangular prism and
tube window are used. Although the rod prism was
selected primarily for packaging reasons, its correction of beam asymmetry makes it worth considering
in other designs.
The short penetration depth of UV light and overlapping excitation and collection spots make the
probe sensitive primarily to superficial fluorophores.
However, the large spot size tends to make the probe
sensitive to a broader range of depths.17 A shallow
probe depth is appropriate for detection of early-stage
adenoma, which forms entirely within the mucosa.
The device was successfully applied to take images
and autofluorescence spectra in vivo in the Min
mouse colon cancer model. The images and spectra
shown indicate sufficient resolution to reproduce earlier research done in our laboratory with excised colon and an in-air combined OCT–LIF instrument.
In vivo capability will allow us to observe changes in
tissue over the term of an experiment with fewer test
animals. The colon was stretched and therefore
thinner than seen in our previous in vitro studies;
however, the relative thickness of the colon can be
determined, and structural details indicating tissue
organization are still visible. Stretching likely
changes the birefringence of the tissue and may account for some observed banding in the OCT images.
The snug fit has an advantage of stabilizing the colon
tissue during the relatively slow measurement, although the looser surrounding connective tissue may
still move. The quality of images cannot be directly
compared with images of the human colon because
the sizes of structural features are so much smaller in
1 January 2004 兾 Vol. 43, No. 1 兾 APPLIED OPTICS
119
the mouse; as an example, the 60 –70-␮m crypts observable in humans are nearly as large as the entire
depth of the mucosa in mice. The primary features
of the emission spectra including peak locations and
absorption dips are consistent with expectations 共Fig.
6兲. The spectra are corrected for instrument sensitivity but are not corrected for tissue transmission in
an attempt to back out intrinsic fluorescence of the
tissue. These raw spectra, including absorption by
blood and other tissue chromophores, have previously
been used successfully to differentiate normal tissue
from adenoma.2,3 Use of live tissue avoids the decay
of fluorophores associated with metabolism, which
depends on time elapsed after resection.3 Stretching
the colon may also squeeze blood from the tissue, resulting in some reduction in fluorescence resorption.
Planned research includes several device optimizations and animal experiments. We will soon build a
second prototype with components verified to be
within design specifications, omitting optical cement
in the junction between the LIF fibers and the rod
prism to reduce transmission loss over time, and
small adjustments made in the diameters of the polyimide endoscope body to aid manufacturability. Automated filter and shutter control will allow effective
data collection at multiple wavelengths and help reduce exposure during nonimaging time. Ongoing
experiments include a full-term observation of the
growth adenomas over the 24-week lifespan of ApcMin
mice, as well as mice with induced colon cancers.
With the histology from these experiments, we will be
able to correlate OCT and fluorescence changes with
tissue changes associated with adenoma formation.
We look forward to potential applications in other
animal models and humans.
6.
7.
8.
9.
10.
11.
12.
This research was supported in part by grants from
the National Institutes of Health Small Animal Imaging Resource 共CA83148兲, Specialized Program of
Research Excellence in Gastrointestinal Cancers,
and the University of Arizona Imaging Small Grant.
The authors thank Gabe Loeb who wrote much of the
spectrometer control software. We also express our
greatest appreciation to Steve Griffin and InnovaQuartz for copious technical assistance and perseverance with the specification and manufacture of the
endoscope tip optical components.
13.
References
17.
1. Cancer Facts and Figures 2003 共American Cancer Society, Atlanta, Ga., 2003兲.
2. R. Richards-Kortum, R. P. Rava, R. E. Petras, M. Fitzmaurice,
M. Sivak, and M. S. Feld, “Spectroscopic diagnosis of colonic
dysplasia,” Photochem. Photobiol. 53, 777–786 共1991兲.
3. K. T. Schomacker, J. K. Frisoli, C. C. Compton, T. J. Flotte,
J. M. Richter, T. F. Deutsch, and N. S. Nishioka, “Ultraviolet
laser-induced fluorescence of colonic polyps,” Gastroenterology
102, 1155–1160 共1992兲.
4. G. I. Zonios, R. M. Cothren, J. T. Arendt, W. Jun, J. Van Dam,
J. M. Crawford, and R. Manoharan, “Morphological model of
human colon tissue fluorescence,” IEEE Trans. Biomed. Eng.
43, 113–122 共1996兲.
5. K. Kobayashi, J. A. Izatt, M. D. Kulkarni, J. Willis, and M. V.
Sivak, Jr., “High-resolution cross-sectional imaging of the gas120
APPLIED OPTICS 兾 Vol. 43, No. 1 兾 1 January 2004
14.
15.
16.
18.
19.
20.
21.
trointestinal tract using optical coherence tomography: preliminary results,” Gastrointest. Endosc. 47, 515–523 共1998兲.
C. Pitris, C. Jesser, S. A. Boppart, D. Stamper, M. E. Brezinski,
and J. G. Fujimoto, “Feasibility of optical coherence tomography for high-resolution imaging of human gastrointestinal
tract malignancies,” J. Gastroenterol. 35, 87–92 共2000兲.
M. V. Sivak, Jr., K. Kobayashi, J. A. Izatt, A. M. Rollins, R.
Ung-Runyawee, A. Chak, R. C. Wong, G. A. Isenberg, and J.
Willis, “High-resolution endoscopic imaging of the GI tract
using optical coherence tomography,” Gastrointest. Endosc.
51, 474 – 479 共2000兲.
R. J. McNichols, A. Gowda, B. A. Bell, R. M. Johnigan, K. H.
Calhoun, and M. Motamedi, “Development of an endoscopic
fluorescence image guided OCT probe for oral cancer detection,” in Biomedical Diagnostic, Guidance, and Surgical-Assist
Systems III, T. Vo-Dinh, W. S. Grundfest, and D. A. Beanaron,
eds., Proc. SPIE 4254, 23–30 共2001兲.
R. V. Kuranov, V. V. Sapozhnikova, N. M. Shakhova, V. M.
Gelikonov, E. V. Zagainova, and S. A. Petrova, “Combined
application of optical methods to increase the information content of optical coherent tomography in diagnostics of neoplastic
processes,” Quantum Electron. 32, 993–998 共2002兲.
V. V. Sapozhnikova, N. M. Shakhova, V. A. Kamensky, R. V.
Kuranov, V. B. Loshenov, and S. A. Petrova, “Complementary
use of optical coherence tomography and 5-aminolevulinic acid
induced fluorescent spectroscopy for diagnosis of neoplastic
processes in cervix and vulva,” in Coherence Domain Optical
Methods and Optical Coherence Tomography in Biomedicine
VII, V. V. Tuchin, J. A. Izatt, and J. G. Fujimoto, eds., Proc.
SPIE 4956, 81– 88 共2003兲.
A. R. Tumlinson, L. P. Hariri, and J. K. Barton, “Miniature
endoscope for a combined OCT-LIF system,” in Coherence Domain Optical Methods and Optical Coherence Tomography in
Biomedicine VII, V. V. Tuchin, J. A. Izatt, and J. G. Fugimoto,
eds., Proc. SPIE 4956, 129 –138 共2003兲.
G. J. Tearney, S. A. Boppart, B. E. Bouma, M. E. Brezinski,
N. J. Weissman, J. F. Southern, and J. G. Fujimoto, “Scanning
single-mode fiber optic catheter-endoscope for optical coherence tomography,” Opt. Lett. 21, 543–545 共1996兲.
B. E. Bouma and G. J. Tearney, “Power-efficient nonreciprocal
interferometer and linear-scanning fiber-optic catheter for optical coherence tomography,” Opt. Lett. 24, 531–533 共1999兲.
Y. T. Pan, H. K. Xie, and G. K. Fedder, “Endoscopic optical
coherence tomography based on a microelectromechanical mirror,” Opt. Lett. 26, 1966 –1968 共2001兲.
U. Utzinger and R. R. Richards-Kortum, “Fiber optic probes for
biomedical optical spectroscopy,” J. Biomed. Opt. 8, 121–147
共2003兲.
N. A. Denisov, “Comparison of competing fiber optic probes for
tissue fluorescence analysis,” in Optical Biopsy and Tissue
Optics, I. J. Bigio, G. J. Mueller, G. J. Puppels, R. W. Steiner,
and K. Svanberg, eds., Proc. SPIE 4161, 234 –243 共2000兲.
T. J. Pfefer, K. T. Schomacker, M. N. Ediger, and N. S. Nishioka, “Multiple-fiber probe design for fluorescence spectroscopy
in tissue,” Appl. Opt. 41, 4712– 4721 共2002兲.
T. J. Pfefer, L. S. Matchette, A. M. Ross, and M. N. Ediger,
“Selective detection of fluorophore layers in turbid media: the
role of fiber-optic probe design,” Opt. Lett. 28, 120 –122 共2003兲.
L. K. Su, K. W. Kinzler, B. Vogelstein, A. C. Preisinger, A. R.
Moser, A. Luongo, K. A. Gould, and W. F. Dove, “Multiple
intestinal neoplasia caused by a mutation in the murine homolog of the APC gene,” Science 256, 668 – 670 共1992兲.
P. C. Chulada, M. B. Thompson, J. F. Mahler, C. M. Doyle,
B. W. Gaul, C. Lee, H. F. Tiano, S. G. Morham, O. Smithies,
and R. Langenbach, “Genetic disruption of Ptgs-1, as well as
Ptgs-2, reduces intestinal tumorigenesis in Min mice,” Cancer
Res. 60, 4705– 4708 共2000兲.
L. H. Colbert, J. M. Davis, D. A. Essig, A. Ghaffar, and E. P.
Mayer, “Exercise and tumor development in a mouse predisposed to multiple intestinal adenomas,” Med. Sci. Sports Exercise 32, 1704 –1708 共2000兲.
22. C. D. Davis, H. Zeng, and J. W. Finley, “Selenium-enriched
broccoli decreases intestinal tumorigenesis in multiple intestinal neoplasia mice,” J. Nutr. 132, 307–309 共2002兲.
23. H. K. Roy, W. J. Karoski, A. Ratashak, and T. C. Smyrk,
“Chemoprevention of intestinal tumorigenesis by nabumetone:
induction of apoptosis and Bcl-2 downregulation,” Br. J. Cancer 84, 1412–1416 共2001兲.
24. E. H. Huang, J. J. Carter, R. L. Whelan, Y. H. Liu, J. O.
Rosenberg, H. Rotterdam, A. M. Schmidt, D. M. Stern, and
K. A. Forde, “Colonoscopy in mice,” Surg. Endosc. 16, 22–24
共2002兲.
25. N. Ramanujam, “Fluorescence spectroscopy of neoplastic and
non-neoplastic tissues,” Neoplasia 2, 89 –117 共2000兲.
26. J. A. Izatt, M. D. Kulkarni, S. Yazdanfar, J. K. Barton, and
A. J. Welch, “In vivo bidirectional color Doppler flow imaging of
picoliter blood volumes using optical coherence tomography,”
Opt. Lett. 22, 1439 –1441 共1997兲.
1 January 2004 兾 Vol. 43, No. 1 兾 APPLIED OPTICS
121