Detection of tumorigenesis in urinary bladder
with optical coherence tomography: optical
characterization of morphological changes
T. -Q. Xie1, 2, M. L. Zeidel2, and Y. -T. Pan1, 2
1
Department of Biomedical Engineering, State University of New York at Stony Brook
HSC-T18, RM 025A, Stony Brook, NY 11794-8181;
2
Department of Medicine, University of Pittsburgh
A1219 Scaife Hall, 3550 Terrace Street, Pittsburgh, PA 15261.
Tuqiang.Xie@sunysb.edu; zeidel@msx.dept-med.pitt.edu; Yingtian.Pan@sunysb.edu
Abstract: Most transitional cell tumorigenesis involves three stages of
subcellular morphological changes: hyperplasia, dysplasia and neoplasia.
Previous studies demonstrated that owing to its high spatial resolution and
intermediate penetration depth, current OCT technology including
endoscopic OCT could delineate the urothelium, submucosa and the upper
muscular layers of the bladder wall. In this paper, we will discuss the
sensitivity and limitations of OCT in diagnosing and staging bladder cancer.
Based on histomorphometric evaluations of nuclear morphology, we
modeled the resultant backscattering changes and the characteristic changes
in OCT image contrast. In the theoretical modeling, we assumed that nuclei
were the primary sources of scattering and were uniformly distributed in the
uroepithelium, and compared with the results of the corresponding prior
OCT measurements. According to our theoretical modeling, normal bladder
shows a thin, uniform and low scattering urothelium, so does an
inflammatory lesion except thickening in the submucosa. Compared with a
normal bladder, a hyperplastic lesion exhibits a thickened, low scattering
urothelium whereas a neoplastic lesion shows a thickened urothelium with
increased backscattering. These results support our previous animal study
that OCT has the potential to differentiate inflammation, hyperplasia, and
neoplasia by quantifying the changes in urothelial thickening and
backscattering. The results also suggest that OCT might not have the
sensitivity to differentiate the subtle morphological changes between
hyperplasia and dysplasia based on minor backscattering differences.
2002 Optical Society of America
OCIS codes: (110.4500) Optical Coherence Tomography; (170.3880) Medical and biological
imaging
References and links
1.
2.
3.
4.
5.
http://www.cancernews.com/category.asp?cat=28&aid=235, “Diagnosis and Treatment of Bladder
Cancer.”
J. M. Schmitt, “Optical Coherence Tomography (OCT): A Review,” IEEE J. Sel. Top. Quantum Electron.
5, 1205-1215 (1999).
A. F. Fercher, K. Mengedoht, W. Werner, “ Eye length measurement by interferometry with partially
coherent light,” Opt. Lett. 13, 186-188 (1988).
D. Huang, E. A. Swanson, C. P. Lin, et al., “Optical coherence tomography,” Science 254, 1178-1181
(1991).
Y. Pan, J. P. Lavelle, S. Meyers, G. Pirtskhalaishvili, M. L. Zeidel, and D. L. Frakas, “Detection of
tumorigenesis in rat bladders with optical coherence tomography,” Med. Phys. 28, 2432-2440 (2001).
#1745 - $15.00 US
(C) 2002 OSA
Received October 10, 2002; Revised November 26, 2002
2 December 2002 / Vol. 10, No. 24 / OPTICS EXPRESS 1431
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
E. A. Swanson, J. A. Izatt, M. R. Hee, D. Huang, C. A. Puliafito, J. S. Schuman, and J. G. Fujimoto, “In
vivo retinal imaging by optical coherence tomography,” Opt. Lett. 18, 1864 (1993).
C. A. Puliafito, M. R. Hee, J. S. Schuman, and J. G. Fujimoto, Optical Coherence Tomography of Ocular
Diseases (SLACK, Thorofare, NJ, 1996).
A. M. Sergeev, V. M. Gelikonov, G. V. Gelikonov, F. I. Feldchtein, K. I. Pravdenko, D. V. Shabanov, N.
D. Gladkova, V. V. Pochinko, V. A. Zhegalov, G. I. Dmitriev, I. R. Vazina, G. A. Petrova, and N. K.
Nikulin, “In vivo optical coherence tomography of human skin microstructure,” Proc. SPIE 2328, 144-150
(1994).
J. Welzel, “Optical coherence tomography in dermatology: a review,” Skin Research and Technology 7, 19 (2001).
B. W. Colston, U. S. Sathyam, L. B. DaSilva, M. J. Everett, P. Stroeve, and L. L. Otis, “Dental OCT,” Opt.
Express 3, 230-238 (1998), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-3-6-230.
G. J. Tearney, M. E. Brezinski, B. E. Bouma, S. A. Boppart, C. Pitris, J. F. Southern, and J. G. Fujimoto,
“In Vivo Endoscopic Optical Biopsy with Optical Coherence Tomography,” Science 276, 2037-2039
(1997).
A. M. Sergeev, V. M. Gelikonov, G. V. Gelikonov, F. I. Feldchtein, R. V. Kuranov, N. D. Gladkova, N. M.
Shakhova, L. B. Snopova, A. V. Shakhov, I. A. Kuznetzova, A. N. Denisenko, V. V. Pochinko, Yu. P.
Chumakov, and O. S. Streltzova, "In vivo endoscopic OCT imaging of precancer and cancer states of
human mucosa," Opt. Express 1, 432-440 (1997), http://epubs.osa.org/oearchive/source/2788.htm.
Johannes F. de Boer, Shyam M. Srinivas, Arash Malekafzali, Zhongping Chen, and J. Stuart Nelson,
“Imaging thermally damaged tissue by polarization sensitive optical coherence tomography,” Opt. Express
3, 212-218 (1998), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-3-6-212.
Gang Yao, Lihong Wang, “Propagation of polarized light in turbid media: simulated animation sequences”,
Opt. Express 7, 198-203 (2000), http://www.opticsexpress.org/abstract.cfm?URI=OPEX-7-5-198.
M. Wojtkowski, A. Kowalczyk, R. Leitgeb, A. F. Fercher, “Full range complex spectral optical coherence
tomography technique in eye imaging,” Opt. Lett. 27, 1415-1417 (2002).
W. Drexler, U. Morgner, F. X. Kartner, C. Pitris, S. A. Boppart, X. D. Li, E. P. Ippen, and J. G. Fujimoto,
“In vivo ultrahigh-resolution optical coherence tomography”, Opt. Lett. 24, 1221-1223 (1999).
Volker Westphal, Siavash Yazdanfar, Andrew M. Rollins, Joseph A. Izatt, “Real-time, high velocityresolution color Doppler optical coherence tomography”, Opt. Lett. 27, 34-36 (2002).
F. I. Feldchtein, V. M. Gelikonov, G. V. Gelikonov, A. M. Sergeev, N. D. Gladkova, A. V. Shakhov, N. M.
Shakhova, L. B. Snopova, A. B. Terent’eva, E. V. Zagainova, Y. P. Chumakov, and I. A. Kuznetzova,
“Endoscopic applications of optical coherence tomography,” Opt. Express 3, 257-270 (1998),
http://www.opticsexpress.org/abstract.cfm?URI=OPEX-3-6-257.
Y. Pan, H. Xie, and G. K. Fedder, “Endoscopic optical coherence tomography based on a
microelectromechanical mirror,” Opt. Lett. 26, 1966-1968 (2001).
T. Xie, H. Xie, G. K. Fedder, M. Zeidel, and Y. Pan, “Endoscopic Optical Coherence Tomography with a
Micromachined Mirror,” 2nd Annual International IEEE-EMBS, Madison, Wisconsin, USA, May 2-4, 208211 (2002).
Tuqiang Xie, Zhigang Li, Mark L. Zeidel, Yingtian Pan, “Optical imaging diagnostics of bladder tissue
with optical coherence tomography”, Proc. SPIE 4609, in print.
Andrew Rollins, Joseph Izatt, Manish Kulkarni, Siavash Yazdanfar, Rujchai Ung-arunyawee, “In vivo
video rate optical coherence tomography”, Opt. Express 3, 219-229 (1998), http://www.opticsexpress.org
/abstract.cfm?URI=OPEX-3-6-219.
Yingtian Pan, Reginald Birngruber, Jurgen Rosperich, and Ralf Engelhardt, “Low-coherence optical
tomography in turbid tissue: theoretical analysis,” Appl. Opt. 34, 6564-6574 (1995).
M. E. Brezinski and J. G. Fujimoto, “ Optical Coherence Tomography: High-Resolution Imaging in
Nontransparent Tissue,” IEEE J. Sel. Top. Quantum Electron. 5, 1185-1192 (1999).
William M. Murphy, Bruce Beckwith, George M. Farrow, Tumors of the kidney, bladder and related
urinary structures, Chapter 2 (Armed Forces Institute of Pathology, Washington, 1994).
J M Schmitt, A Knuttel, M Yadlowsky and M A Bckhause, “Optical-coherence tomography of a dense
tissue: statistics of attention of backscattering,” Phys. Med. Biol. 39,1705-1720 (1993).
J. M. Schmitt and A. Knuttel, “Model of optical coherence tomography of heterogeneous tissue,” J. Opt.
Soc. Am. A 14, 1231-1242 (1997).
M. Kowalevicz, T. Ko, I. Hartl, J. G. Fujimoto, M. Pollnau, and R. P. Salathe, “Ultrahigh resolution optical
coherence tomography using a superluminescent light source,” Opt. Express 10, 349-353, 2002.
Gang Yao and Lihong V Wang, “Monte Carlo simulation of an optical coherence tomography signal in
homogeneous turbid media,” Phys. Med. Biol. 44, 2307-2320 (1999).
M. Kriegmair, R. Baumgartner, R. Knuechel, H. Stepp, F. Hofstaedter and A. Hofstetter, "Detection of
early bladder cancer by 5-aminolevulinic acid induced porphyrin fluorescence," J. of Urology 155, 105-110
(1996).
F. Koenig, F. J. McGovern, R. Larne, H. Enquist, K. T. Schomacker and T. F. Deutsch, "Diagnosis of
bladder carcinoma using protoporphyrin IX fluorescence induced by 5-aminolaevulinic acid," BJU
International 83, 129-135 (1999).
#1745 - $15.00 US
(C) 2002 OSA
Received October 10, 2002; Revised November 26, 2002
2 December 2002 / Vol. 10, No. 24 / OPTICS EXPRESS 1432
1. Introduction
Clinical statistics indicate that bladder cancer is the fifth most common cancer and the twelfth
leading cause of cancer death in the US1. Bladder cancer is curable if detected and treated
prior to invasion in the underlying bladder wall. Therefore, an earlier and more precise
diagnosis of bladder cancer is critical to eradicating the disease, and understanding the
pathological mutation of the disease on the microscopic level could lead to a better diagnosis
because bladder carcinoma originates in the thin (20-200 µm) basal cell layer of the
uroepithelium. Staging the spread or extent of invasion of cancerous urothelial cells into the
underlying bladder wall is also important in helping the urologists to design the best treatment
strategy. However, because of limitations of resolution, current detection methods, e.g., urine
cytology, intravenous x-ray, MRI, and ultrasound fail to provide sufficient sensitivity or
specificity2-6 in predicting the prognosis of early bladder cancers and staging their invasions.
Cystoscopy, although commonly used in urological examination of superficial tumors (e.g.,
carcinomas in situ), is always followed by random biopsy for a conclusive diagnosis because
of lacks of depth resolution. Accordingly, new approaches that can instantaneously provide
cross-sectional images of bladder morphologies and their alternations (e.g., tumorigenesis) at
close to cellular resolution would substantially enhance the diagnostic sensitivity and
specificity of current cystoscopic approaches and result in significant therapeutic benefits.
Since its first introduction to imaging the eye in early 19903,4, optical coherence
tomography (OCT) has found widespread applications in diagnosing diseases in various
biological tissues7 such as human skin8, 9, tooth10, blood vessels, gastrointestinal tracts,
respiratory tracts, and genitourinary tracts11, 12. In recent years, significant technological
advances including polarization-sensitive OCT13, 14, spectral OCT15, ultra-high-resolution
OCT16, and Doppler OCT17, have been made to improve image resolution and provide more
specific diagnosis of physiological and functional information of biological tissue.
Furthermore, because OCT is a fiber optically based light scanning imaging technique it can
be integrated with endoscopic catheters to allow for noninvasive or minimally invasive in
vivo imaging diagnosis of intraluminal tracts. Implementations of real-time endoscopic OCT
using a rotary joint, a PZT transducer and a MEMS micromirror have been reported11, 18-20,
showing a great promise to leverage the diagnostic capability of current endoscopic
modalities.
Both ex vivo and in vivo OCT imaging of urinary bladder has been reported8, 16, 21;
however, there has been no systematic study that correlates the micro morphology imaged by
OCT with that provided by the corresponding histology. Such a study is needed to examine
the utility and limitations of OCT in diagnosing and staging bladder cancers. In our previous
paper5, we reported the first systematic OCT study of the course of bladder tumorigenesis in a
well-characterized model in which Fisher rats were exposed to methyl-nitroso-urea (MNU),
followed with cystoscope-like surface imaging, OCT and histology. The results suggested the
potential of OCT for detection of bladder inflammatory lesions and early urothelial
abnormalities by analyzing the variations in urothelial thickening and backscattering.
However, further analytical study is needed to characterize the micro morphological changes
at different stages of tumorigenesis (e.g., hyperplasia, dysplasia and neoplasia) with respect to
the resultant changes in the OCT signature, i.e., backscattering, so that we can better predict
the sensitivity and limitations of OCT in staging transitional cell tumors. In the present
studies, we injected water, blood and intralipid into the submucosa of rat bladders to simulate
different types of inflammatory lesions (e.g., edema, vasocongestion, inflammatory infiltrate)
and characterized the resultant OCT contrast changes. We next modeled the optical
characteristics and the resultant OCT signal changes based on the histomorphometric
evaluations of subcellular morphological changes of urothelial cells at the three stages (i.e.,
hyperplasia, dysplasia and neoplasia) during tumorigenesis. We then compared the modeling
results with the OCT measurements for those samples whose histological evaluations were
#1745 - $15.00 US
(C) 2002 OSA
Received October 10, 2002; Revised November 26, 2002
2 December 2002 / Vol. 10, No. 24 / OPTICS EXPRESS 1433
previously harvested for optical modeling. This comparative study is important to characterize
the OCT image contrast as a result of tumorigenesis and analyze the sensitivity of OCT for
diagnosis and staging of bladder cancer.
2. Methods
2.1 High-Performance OCT Imaging System
The principle of OCT is similar to that of ultrasound imaging. A schematic diagram of our
OCT system used to perform 2D cross-sectional images presented in this study is shown in
Fig. 1, which is based on a fiber-optic Michelson interferometer. This technique performs
high-resolution optical ranging or tomographic measurement by taking advantage of the short
temporal coherence of a broadband light source. A high-brightness, broadband light source
was used for illumination. Its pigtailed output power I is 13mW, central wavelength λ is
1320nm, and half-maximum-full-width (HMFW) spectral bandwidth ∆λ is 77nm, thus
yielding a coherence length LC of roughly 10µm2, 23. The broadband light was split equally
into the reference and the sample arms of the fiberoptic Michelson interferometer. Because a
low-coherence source was used, the recombined light in the detection fiber only coherently
interfered when the pathlengths in the sample and reference arms were matched to within the
short coherence length, Lc≈10µm. Thus, moving the reference mirror permitted tomographic
determination of the pathlength-resolved distribution of the interference modulation of light
from the reflecting interfaces within the biological tissue placed in the sample arm. This
offers a unique advantage of OCT over any other optical imaging modality in that it
circumvents the need to scan a bulky microscopic lens to accomplish high-resolution imaging.
The axial reflectance profile, i.e., the envelope of the interferometric signal was detected at
high sensitivity by using optical heterodyne detection, locking in the Doppler frequency
shifted signal (fD = 2v/λ, v is the speed for reference mirror scan). This signal was bandpass
filtered and envelope demodulated prior to feeding to a PC for image display. Scanning the
reference mirror at stable and high speed yields a constant Doppler frequency for optical
heterodyne detection, and is thus critical to real-time high-fidelity OCT imaging. In our
system, this was accomplished by using a double-pass grating-lens based optical delay line, a
techniques used for fs laser Fourier-transform pulse shaping22. With proper settings of all the
components, optical ranging up to 3mm was achieved at over 1kHz, thus allowing 2D OCT
imaging at almost 4-8 frames/s.
In the sample arm, the light beam existing the fiber was collimated, deflected by a twoaxis servo scanner, and then focused on the tissue specimen under examination with a
microscopic objective (4x, NA0.1). A red laser diode (λ = 670 nm) along with a color camera
was used to align the lateral position of the tissue specimen under examination; whereas the
axial focusing of the incident beam was adjusted by the preset reference mirror position as
precisely as less than 5 micrometers, taking advantage of the short source coherence length.
A 2D cross-sectional image was produced by scanning the incident light beam across the
tissue with a lateral servo scanner each time after the sequential reflectivity profile in the
longitudinal direction was taken. The axial (∆z) and lateral (∆r) resolutions of the OCT
system are determined by the coherence length and the microscopic lens used in the probe
beam, respectively. The axial resolution (∆z) is 23, 24
∆z = LC = (2 ln 2 π ) ⋅ (λ
2
∆λ )
(1)
The lateral or transverse resolution is determined by the focused spot size in analogy with
conventional microscopy and is
∆r = 2 λ0 πNA = 4λ0 f πφ
#1745 - $15.00 US
(C) 2002 OSA
(2)
Received October 10, 2002; Revised November 26, 2002
2 December 2002 / Vol. 10, No. 24 / OPTICS EXPRESS 1434
where NA = φ/2f is the numerical aperture of the microscopic lens, f is the focal length of the
microscopic lens and φ is the spot size of the light beams exiting the fiber optic collimator.
For the OCT setup used in this study, they were roughly 10µm, respectively.
Phase
Control
Delay Line
Galvanometer
Mirror
BBS
1320 nm
CM
95:5
LD
670 nm
50:50
CM
PD
Grating
Lateral
Scanning
Mirror
Scan Head
X
z
Analog Signal Processing
Computer
Fig. 1. Schematic diagram of the fiber optic OCT system. BBS: broadband light source; LD:
aiming laser diode; PD: photo diode; CM: fiber-optic collimator. High-speed reference mirror
scanning is grating-lens delay line.
2.2 Optical modeling
To characterize OCT with histology for micro morphological changes induced by
tumorigenesis, a controllable bladder tumor model was used by way of instillation of proper
doses of a chemical carcinogen, MNU, into the bladder of Fisher rats. At different time points,
e.g., at week 20, 24, 30, 36, the intact rat bladders were harvested, properly stretched and
mounted onto φ10mm ring holders filled with 37°C modified Ringer's buffer solution. The
apical or urothelial side of the bladder was faced up for cystoscope-like surface imaging,
OCT, and H&E stained histology. The lateral positions of the OCT scan were precisely
tattooed in Indian ink with the help of a red aiming laser and color CCD camera to guide later
histological sectioning and light microscopy.
For the simulation study of different types of bladder inflammatory lesions, 3 groups of a
total of 9 fresh normal rat bladder samples were mounted on ring holders to allow OCT
imaging and the section scanned by OCT were pinned by a pair of 32 gauge needles as
landmarks. Then, for group 1 and group 2, approximately 5 µl 0.9% saline and fresh whole
blood previously taken from the same animal were injected into the submucosal layer of the
sections previously scanned by OCT to simulate edema and severe edema with vasodilation,
respectively. For group 3, the same amount of 10% intralipid was injected into submucosa to
simulate accumulated inflammatory cells and other types of inflammatory infiltrate (e.g.,
necrosis, fibrosis) with increased scattering. Microinjection was performed under a stereo
dissecting microscope using 32 gauge needles. A comparison between OCT images prior to
and post microinjection allowed us to characterize OCT images for the changes of
submucosal morphology and the resultant optical properties of bladder inflammatory lesions.
Clinical diagnosis of malignancy presently relies on high-resolution pathological analysis
of subcellular morphology, e.g., increase of nuclear to cytoplastic ratio (NC ratio), and loss of
polarity25. Because of technological limitations, current OCT (endoscopic OCT, in particular)
does not provide sufficient resolution to image these subcellular changes in vivo.
#1745 - $15.00 US
(C) 2002 OSA
Received October 10, 2002; Revised November 26, 2002
2 December 2002 / Vol. 10, No. 24 / OPTICS EXPRESS 1435
Nevertheless, the results of our previous study5, 21 suggested that the change in urothelial
backscattering induced by nuclear morphology (e.g., increase in NC ratio, loss of polarity)
detected by OCT might provide a potential method to differentiate between urothelial
hyperplasia (precancer) and neoplasia (cancer).
Unfortunately, complicated wave optical effects (e.g. multiple scattering and speckle
effects) in random medium such as biological tissue have hindered theoretical modeling of
OCT; presently, there is no analytical or numerical model that can effectively analyze the
OCT image contrast with respect to the histological evaluations. To tackle the challenge, we
explored this complicated problem by combining analytical modeling with experimental
measurements. For each tumor sample, multiple OCT scans were performed and followed by
histological sections on the same areas. The H&E stained histological sections were
photographed by a 4k×3k-pixel color digital CCD camera. Once hyperplastic, dysplastic and
neoplastic lesions in these samples were identified, image processing was applied to the areas
of the urothelium with pathological alternations. In the optical modeling of cellular scattering,
we neglected the contributions of other scattering sources, e.g., mitochondria, cytoplastic
contents and intercellular boundaries, and assumed that nuclei were the primary scattering
sources of urothelial cells. Then, the urothelium was optically simplified as a 2D matrix
consisting of scattering centers (nuclei) with refractive index of nN = 1.52 surrounded by a
nonscattering solution with nC=1.38. The nuclei were assumed to be spherical and their sizes
were counted to yield the mean nuclear size (dN) and ‘2D’ volumetric density ρs for scattering
calculation. For mathematic simplicity, we further assumed all scattering centers were
uniformly distributed, and the simplified urothelial model became a uniform 3D scattering
matrix with equal particle size of dN and refractive index of nN. The 3D volumetric particle
density ρv were calculated according to the relation,
ρv = 1.33ρs3/2
(3)
Thus, based on the measured mean particle size dN and volumetric density ρs, the scattering
characteristics, e.g., scattering coefficient µs, scattering anisotropy g or backscattering
coefficient µb were calculated according to Mie-scattering algorithms. Based on our previous
study, the interferometric signal detected by OCT was approximately given by 23, 26-28
I%d ( L r ) = 2 I sI r ⋅
R ( L s ) ⊗ C( L s )
(4)
where Ir is light intensity in the reference arm, Is is the light incident on the biological tissue.
[
]
C(Ls) = exp − 4 ( L s L c ) 2 cos k L s is low-coherence function, and R(Ls) is the
pathlength-resolved reflectance. Urothelium is a thin and relatively low-scattering tissue
(except large papillary neoplastic lesions). Assuming that light scattering within the urothelial
layer is homogeneous and within the single scattering regime; then, we could approximate the
relations, Ls ≅ 2ncz (z is geometric depth), R(Ls) ≅ µbe-µs2z. Thus, Eq.(4) can be simplified as
~
I d ( z ) = kI 0 µb e − µ S z
∑
e − 4( ∆L Lc ) cos k ∆L
2
∆L ≤ LC
(5)
The modular term relates to the speckle effect, i.e., the summation of the backscattered light
fields from a local volume (∆L≤Lc) containing a large collection of particles (nuclei), each of
which has a probability of scattering light at different angles, polarization directions, and at
different delays or phase shifts depending on the geometric distribution. Therefore, we can
neglect the speckle effects and characterize the ‘speckle-free’ OCT image contrast changes on
lesions during tumorigenesis by evaluating the first term µb e− µ S z base on the measured
histological distribution and the model analysis stated above.
#1745 - $15.00 US
(C) 2002 OSA
Received October 10, 2002; Revised November 26, 2002
2 December 2002 / Vol. 10, No. 24 / OPTICS EXPRESS 1436
3. Results
As previously reported5, 14, early acute chemical cystitis following MNU instillation involves
injuries of urothelium, causing leakage of urine constituents through a defective bladder
permeability barrier which begins a process of chronic inflammation of the underlying
muscular layers and in turn causes fibrosis and urothelial malignancies. This protocol
provides a controllable animal model to permit systematic study of bladder diseases such as
inflammatory lesions, cystitis, and transitional cell tumorigenesis. For acute injury to the
bladder, our previous OCT study revealed that based on its ability to identify urothelium,
submucosa and muscles as well as the backscattering changes in these layers, OCT was able
to detect urothelial denudation, submucosal edema, vasodilation, and infiltrate as evidenced
by the corresponding histology5. However, because of complications in inflammatory changes
which might involve changes in scattering, absorption or both and even the positioning of
these changes, a conclusive diagnosis is sometimes difficult and an in-depth understanding is
highly demanded. To study this problem, we simulated different types of inflammatory
lesions by injecting saline, whole blood and intralipid into the submucosa of the rat bladders.
The results are presented in Fig. 2.
Normal
Saline
U
U
SM
SM
M
M
A
U
B
Intralipid
U
SM
SM
M
M
Blood
C
D
Fig. 2. OCT images of a normal rabbit bladder (A) and rabbit bladder samples injected with
saline (B), blood (C) and intralipid (D). U: urothelium, SM: submucosa, M: muscular layers.
Panel A is the OCT image of a normal rat bladder. The three layers were clearly
delineated based on their scattering difference and structural characteristics: the loosely
stretched urothelium (U) is shown as a low scattering thin layer (∼55µm thick), the
submucosa (SM) is shown as a homogeneous high scattering layer (∼190µm thick). Because
of scattering induced light lose, the ∼240µm thick muscular layers (M) is shown as less bright
than SM but can be identified by the boundaries of bifurcated collagen bundles. Panel B is the
OCT image following ∼5µl 0.9% saline injection to simulate early edema with fluid buildup.
Because of low viscosity, some saline might flow out of the bladder wall through the injection
hole; however, minor swelling in SM is clearly visible (∼280µm thick). In addition to the dark
stripes filled with saline, the overall SM is slightly less scattering than that the normal SM in
Panel A possibly resulted from saline perfusion. Like Panel A, the underlying bladder wall is
#1745 - $15.00 US
(C) 2002 OSA
Received October 10, 2002; Revised November 26, 2002
2 December 2002 / Vol. 10, No. 24 / OPTICS EXPRESS 1437
clearly visible. Panel C is the OCT image following ∼5µl whole bladder injection to simulate
SM edema with vasodilation. The injection hole is not shown in the cross section; however,
moderate swelling in SM is clearly visible (∼350µm thick). Compared with Panels A and B, a
substantial decrease of backscattering in SM is observed following microinjection of blood;
and as a result, the underlying bladder morphology disappears. Panel D is the OCT image
following ∼5µl 10% intralipid injection to simulate accumulated inflammatory cells and other
types of inflammatory lesions in SM resulting in increased scattering (e.g., necrosis fibrosis).
Swelling in SM is detected (∼310µm thick). Compared with Panel C, a localized increase of
backscattering in the injected SM is observed following microinjection of intralipid. As a
result, the underlying bladder morphology became less clear than Panels A and B.
Overall, the simulation results presented above support our previous study5 in that
different types of inflammatory lesions can be identified by the scattering and absorption
changes associated with the morphological changes. Because fluid buildup presented in early
edema is less turbid or scattering, the overall bladder structure including the underlying
muscular layers is clearly delineated by OCT as seen in Panel B. Blood at 1.3µm is of both
high scattering and absorption (µs>20mm-1, g>0.95, µa>0.5mm-1). Because of its highly
forward scattering and high absorption nature, vasodilation may appear as less bright and
absorptive, causing the underlying bladder structure invisible as has been seen in Panel C.
Intralipid (µs≈10mm-1) is less scattering than whole blood but its scattering anisotropy is
lower (g < 0.7). As a result, it shows a brighter SM in the injected area with missing
underlying structure because of scattering loss (in the above layer). Inflammatory lesions with
accumulated inflammatory cells and fibrosis, because both have higher µs and g, tend to show
both higher backscattering and high attenuation in OCT images.
a) normal
c) dysplatic
b) hyperplastic
d) neoplastic
Fig. 3. Histologic pictures of normal, hyperplastic, dysplastic, and neoplastic urothelial cells.
Figure 3 shows a group of H&E stained histological pictures of rat bladders at four
critical stages of tumorigenesis: Panels A-D are normal, hyperplastic, dysplastic and
neoplastic urothelia, respectively. By comparing the nuclear morphology, we can see that the
#1745 - $15.00 US
(C) 2002 OSA
Received October 10, 2002; Revised November 26, 2002
2 December 2002 / Vol. 10, No. 24 / OPTICS EXPRESS 1438
hyperplastic lesion (B) depicts a similar distribution to that of normal urothelium (A) except
the increase in cell layers or depth; whereas the dysplastic and neoplastic lesions (C and D)
show not only increase in urothelial thickness but also increase in nuclear density and loss of
polarity. In addition, Panel D also depicts heavy vascularization, which is typical for
transitional cell cancers. Further image analysis of nuclear morphology in these pictures leads
to the size of nuclei and their volumetric distributions as presented in Table 1.
Table 1 reveals that from normal to neoplastic urothelia, the nuclear size decreases
slightly from 7.25 µm to 6.14 µm whereas the nuclear density increases drastically from
0.0011/µm to 0.0025/µm resulting in an increased nuclear to cytoplastic ratio. Using the
measured data Table 1 we can calculate the corresponding backscattering and scattering
coefficients µb and µs according Mie theory of scattering and the resultant OCT signals using
Eq. (5).
Table 1: Histological evaluations of nuclear morphology
urothelia
parameters
nuclear size (µm)
hyperplastic
7.52
6.9
6.75
6.14
0.0011
0.0012
0.0025
7.96*10-4
volumetric density (1/mm3)
dysplastic
neoplastic
Normal
Figure 4 shows the simulated OCT signal changes according to Eq. (5). The result
suggests that because of insignificant difference between curves A and B ( ≤ 15%), OCT may
not have the sensitivity to differentiate the subtle micro morphological changes between
hyperplasia and dysplasia. However, due to high NC ratio and the resultant backscattering
increase OCT is able to differentiate neoplasia (cancer) from hyperplasia (precancer) with
over 50% sensitivity. These results are in agreement with our previous studies: by comparing
backscattering change, OCT was able to differentiate hyperplasia and neoplasia or precancer
with TCC. Then, by quantifying urothelial thickening, OCT can differentiate precancer
(hyperplasia and dysplasia) from normal urothelium.
9
8
Backscattering (Arb )
7
6
Normal
Normal
Hyperplasia
Hyperplasia
Dysplasia
Dysplasia
Neoplasia
Neoplasia
5
4
3
2
1
0
10
20
30
40
50
60
70
80
90
100
Depth (µm)
Fig. 4. Calculated results of backscattering changes as a function of nuclear morphology (e.g.,
size, density depicted in Fig. 3).
#1745 - $15.00 US
(C) 2002 OSA
Received October 10, 2002; Revised November 26, 2002
2 December 2002 / Vol. 10, No. 24 / OPTICS EXPRESS 1439
Figure 5 presents a pair of 2D OCT images of a hyperplastic lesion and a neoplastic
lesion along with their histological pictures to validate the theoretical modeling. Two bladder
samples were obtained at weeks 28 and 36 following MNU instillation. Although the bladder
samples might have shrunk during histological processing, the overall micro morphology,
e.g., U, SM and M provided by OCT correlates well with the histology. The thickened
urothelia are indicated by U’. Normal urothelia under stretch were roughly 40-45µm thick.
The hyperplastic lesion U’ in Panel A was 250µm or about 5 times thicker than the normal
urothelium and the neoplastic lesion U’ in panel B was 900 µm or almost 20 times thicker the
normal urothelium, both of which were substantially thickened. It is also demonstrated that
the backscattering increases slightly from normal to hyperplasia (≤25%) but noticeably from
hyperplasia to neoplasia (≥70%) in Fig. 5.
‘
‘
250 µm
250 µm
U’
‘
A) OCT image and histology of bladder with hyperplasia
B) OCT image and histology of bladder with neoplasia
Fig. 5. Comparisons of hyperplastic and neoplastic rat bladders imaged by OCT with histology.
U: normal urothelium, SM: submucosa, M: muscular layer, U’: diseased urothelium.
To further quantify the backscattering changes, image processing was applied to extract
the A-scans or vertical intensity profiles of OCT measurements both at the center of the
urothelial lesions and in the surrounding normal urothelia for comparison. About 10
consequential A-scans were averaged laterally to reduce speckle noise in the calculation, and
the results are shown in Fig. 6 in which the measured OCT signals related to backscattering
are presented in arbitrary unit.
A comparison between these two groups of A-scans reveals that the measured OCT
signal levels for normal urothelia were about 40 counts whereas those for hyperplastic and
neoplastic lesions were 50 and 85 counts, respectively. In other words, compared with normal
urothelium, there was less than 20% increase for the hyperplastic lesion and over 60%
increase for neoplastic lesions, which is approximately in agreement to our theoretical
modeling indicated in Eq. (5) and Fig. 4. The results support our previous OCT study and
reveal that the normal bladder, inflammation, hyperplasia and neoplasia can be diagnosed by
OCT. Inflammatory lesions can be differentiated based on submucosal swelling and the
changes in optical turbidity (low scattering for fluid buildup and high scattering for
accumulated inflammatory cells or fibrosis) and absorption (vasodilation). Urothelial
hyperplasia (precancer) and neoplasia (TCC) can be differentiated from normal urothelium
based on detection of the combined urothelial thickening and increase in backscattering.
#1745 - $15.00 US
(C) 2002 OSA
Received October 10, 2002; Revised November 26, 2002
2 December 2002 / Vol. 10, No. 24 / OPTICS EXPRESS 1440
SM
240 µ m
50
40
U
70 µ m
U 0.25 mm
A: A- scan of OCT image with normal urothelium
B: A- scan of OCT image with hyperplastic lesion
85
40
U
75µm
U
0.90 mm
C: A- scan of OCT image with normal urothelium
D: A-scan of OCT image with neoplastic lesion
Fig. 6. A-scans on the normal, hyperplastic and neoplastic regions acquired from the OCT
images in Fig. 5. Consecutive A-scans were averaged to reduce speckle noises.
4. Discussion
In this study, we simulated different types of bladder inflammatory lesions using normal rat
bladders by microinjection of saline, whole bladder and intralipid. Based on this experimental
model, we characterized the OCT signatures with respect to edema (e.g., fluid buildup),
vasodilation and accumulated inflammatory cells, necrosis or cystitis. Based on the
assumption that nuclei were the primary scattering contributors in the urothelial cells, we
proposed a semi-analytical model to characterize the changes of optical properties (e.g.,
scattering and absorption) and OCT signatures due to micro morphological changes at
different stages of tumorigenesis, and compared the simulation results with the corresponding
OCT images. Both simulation results were in agreement with the experiments, suggesting that
OCT has the potential to provide detailed morphological changes below the bladder surface to
detect submucosal inflammation, urothelial precancer and early cancer by analyzing
uroepithelial thickening and backscattering. The simulation results presented in Fig. 2 for
bladder inflammatory lesions confirm the identifications or analyses of our previous study5,
i.e., different types of submucosal inflammations could be discriminated by the variations of
sunmucosal scattering and absorption. Early edema, modeled by saline injection, was
characterized by an almost transparent, thickened submucosa with clear underlying bladder
wall in OCT (panel B); severe edema with vasodilation, modeled by whole blood injection,
was characterized by a dark, thickened submucosa with missing underlying bladder wall in
OCT (panel C) because of the high-absorption and forward-scattering nature of blood at
1.3µm; inflammatory lesions (e.g., accumulated inflammatory cells, necrosis and fibrosis),
simulated by intralipid injection, was characterized by a bright, thickened submucosa with
#1745 - $15.00 US
(C) 2002 OSA
Received October 10, 2002; Revised November 26, 2002
2 December 2002 / Vol. 10, No. 24 / OPTICS EXPRESS 1441
missing underlying bladder wall in OCT (panel D). This is because with a beginning of
chronic inflammation of the muscular layers, processes of sclerosis and fibrosis prevail in the
connective tissue of the subepithelial and submucosal layers and the scattering from these
structures increases. As the result, there is an increased brightness in these structures and in
turn results in shadows below them as indicated by the arrow in Panel D. In all these lesions, a
common feature was that the urothelium remained normal, i.e., no observation of abnormal
thickening or obvious increase of backscattering (except focal denudations of urothelium).
Unlike other biomedical optical imaging techniques (i.e., time-resolved imaging), OCT
provides microscopic images of turbid biological tissues. Current tissue optics derived within
the framework of photon migration fails to provide useful description of OCT in high
scattering regime. Furthermore, because of insufficient resolution for subcellular imaging and
complications of other wave optical effects such as speckles, an effective modeling of OCT
contrast and resolution in scattering tissue that can relate to tissue microscopic phenomena
still remains a prevailing challenge in OCT study. To tackle this challenge, we proposed a
semi-analytical model to analyze the relation of OCT contrast change as a result of the
subcellular morphological changes induced by tumorigenesis. In this model, we assumed that
nuclei were the predominant scattering sources of urothelial cells. Then, based on
histomorphometric evaluations of nuclear morphology of urothelial cells (Fig. 2) at different
stages of tumor development, we measured the mean scattering sizes and volumetric densities
of these nuclei and the corresponding scattering characteristics (e.g., µs, g or µb) and
calculated the resultant OCT contrast changes in the urothelium (Eq .(4)). The model analysis
revealed that both nuclear density and size might vary with urothelial abnormalities, e.g., from
normal urothelium to hyperplasia, dysplasia and neoplasia; nevertheless, our assessment
showed that the changes in nuclear density was the major factor affecting the scattering
performance in the urothelium and the measured OCT contrast. As has been demonstrated in
Fig. 4, our optical modeling revealed a less than 15% increase in backscattering from normal
urothelium to hyperplasia but an over 50% increase from precancer (hyperplasia, dysplasia) to
TCC (neoplasia). Experiments presented in Fig. 5 and Fig. 6 showed that the measured OCT
signals increased less than 25% at hyperplastic stage and over 70% at neoplastic stage. These
model analyses confirmed the results of our previous animal study and demonstrated the
capability of OCT to discriminate neoplastic urothelium from normal or precancerous
urothelium based on their backscattering changes. Our model analysis also suggested that
OCT might not have the sensitivity to resolve the subtle changes between hyperplasia and
dysplasia based on the insignificant backscattering changes. The results demonstrated that
because of resolution limitations, OCT was unable to image subcellular details. However,
based on our study, this technology has the resolution to delineate cross-sectional bladder
morphology (e.g., U, SM and upper MS) and has the sensitivity to image the local
backscattering changes that reflect the subcellular and nuclear morphological changes induced
by tumorigenesis. Therefore, by imaging the location of the lesions and urothelial thickening,
OCT is potentially able to differentiate submucosal inflammations and urothelial
abnormalities; by further analyzing the backscattering changes and location, OCT is
potentially able to differentiating the types of inflammatory lesion as well as precancers (e.g.,
hyperplasia and dysplasia) and cancers (e.g., neoplasia) in the urothelium. Because the
identification is based on local backscattering changes, OCT might fail to differentiate some
complicated lesions. For instance, a lesion of severe edema with decreased SM scattering
(e.g., vasodilation) may appear as a hyperplastic lesion; and an inflammatory lesion (e.g.,
necrosis or fibrosis) with denuded urothelium may appear as a neoplastic lesion in OCT
image. Our experimental studies5, 21 suggest that the diagnostic sensitivity to differentiate
these types of lesions can be greatly enhanced by comparing them with the morphology of the
adjacent normal bladder wall.
Our theoretical model was semi-analytical and relied on histomorphometric evaluations
as input, which is not readily available in most cases. It was mathematically simplified based
#1745 - $15.00 US
(C) 2002 OSA
Received October 10, 2002; Revised November 26, 2002
2 December 2002 / Vol. 10, No. 24 / OPTICS EXPRESS 1442
on assumptions that nuclei were the primary scattering contributors, and were spherical,
homogeneously distributed and in the single-scattering regime. Nevertheless, the results
revealed that this model provided useful explanations to analyze the effects of subcellular
morphological changes of tumorigenesis on the resultant OCT signature or contrast as a result
of backscattering changes. Further improvement of the theoretical model will consider the
heterogeneous nature of the urothelium (e.g., umbrella, intermediate and basal cell layers),
loss of polarity (e.g., orientation of elongated nuclei), blood scattering and absorption due to
microvascularization in the neoplastic urothelium to provide more precise analyses of these
changes on OCT contrast.
The experiments presented in this study were ex vivo because rats, the only controllable
animal model to generate transitional cell cancers by MNU installation, are too small to
perform cystoscopy. However, we have developed and refined OCT endoscopes based on
microelectromechanical system (MEMS) technology19, 21 that have demonstrated image
resolution and contrast close to the bench-top setup used in this study. We are in the process
to launch in vivo study for the detection of TCCs and carcinomas in situ based on the
expertise we have accumulated in the current systematical studies using this controllable
animal tumor model. On the other hand, recent technological advances in ultra-broadband fs
laser technology have permitted subcellular OCT at 1-3µm resolution16, 24, thus immensely
enhancing the sensitivity and specificity of OCT for the diagnoses of bladder cancers and
other types of epithelial cancers. These OCT technological developments and clinical
applications have demonstrated the potential of endoscopic OCT for noninvasive or
minimally invasive and instantaneous ‘optical biopsy’ or ‘optically guided biopsy’ to
diagnose and stage early-stage cancers. Recent studies have showed that 5-aminolevulinic
acid (5-ALA) induced fluorescence endoscopy has demonstrated the potential to enhance the
diagnosis of malignant/dysplastic bladder lesions30-31; and our recent ex vivo study suggests
that a new systoscopic suite guiding OCT with a 5-ALA fluorescence scope can drastically
enhance the diagnostic specificity for precancerous lesions and reduce the time needed to scan
the entire bladder (submitted to J. of Urology).
5. Conclusions
In summary, OCT has demonstrated the capability to delineate urinary bladder morphology,
e.g., urothelium, submucosa and upper muscular layers as evidenced by the corresponding
histology. Our model study confirms that OCT may be able to identify different types of
submucosal inflammatory lesions based on the difference of scattering and absorption patterns
in these lesions. Based on histomorphometric evaluation of nuclear morphology, the proposed
semi-analytical model predicts a minor difference (<25%) in backscattering between normal
and hyperplastic urothelia and a substantial increase (70%) between hyperplastic and
neoplastic urothelia, which is approximately in agreement with OCT measurements. These
results demonstrate that OCT has the potential to differentiate normal bladder, inflammatory
lesion, urothelial precancer and TCC by analyzing the location of the lesion, urothelial
thickening and backscattering. Overall, OCT has the potential to serve as useful biopsyguiding tool for noninvasive or minimally invasive diagnosis of bladder tumors to reduce
random, negative biopsies.
Acknowledgements
This research was supported in part by grants from the Whitaker Foundation (Y. P.) and the
National Institutes of Health (Y. P., M, Z.). Special thanks to Susan Meyers and R. Ramage in
the Department of Medicine, University of Pittsburgh for handling the animal studies. Please
address future correspondence to Y. -T. Pan by email at yingtian.pan@sunysb.edu.
#1745 - $15.00 US
(C) 2002 OSA
Received October 10, 2002; Revised November 26, 2002
2 December 2002 / Vol. 10, No. 24 / OPTICS EXPRESS 1443