Ultrastructural alterations in field carcinogenesis measured by ...€¦ · Ultrastructural...

Ultrastructural alterations in fieldcarcinogenesis measured by enhancedbackscattering spectroscopy

Andrew J. RadosevichNikhil N. MutyalJi YiYolanda Stypula-CyrusJeremy D. RogersMichael J. GoldbergLaura K. BianchiShailesh BajajHemant K. RoyVadim Backman

Downloaded From: http://spiedigitallibrary.org/ on 10/03/2013 Terms of Use: http://spiedl.org/terms

Ultrastructural alterations in field carcinogenesismeasured by enhanced backscattering spectroscopy

Andrew J. Radosevich,a Nikhil N. Mutyal,a Ji Yi,a Yolanda Stypula-Cyrus,a Jeremy D. Rogers,a Michael J. Goldberg,cLaura K. Bianchi,c Shailesh Bajaj,c Hemant K. Roy,b and Vadim Backmana

aNorthwestern University, Department of Biomedical Engineering, Tech E310, 2145 Sheridan Road, Evanston, Illinois 60208bBoston Medical Center, Department of Medicine, Boston, Massachusetts 02118cNorthShore University Healthsystems, Department of Internal Medicine, Evanston, Illinois 60201

Abstract. Optical characterization of biological tissue in field carcinogenesis offers a method with which to studythe mechanisms behind early cancer development and the potential to perform clinical diagnosis. Previously, low-coherence enhanced backscattering spectroscopy (LEBS) has demonstrated the ability to discriminate between nor-mal and diseased organs based on measurements of histologically normal-appearing tissue in the field of colorectal(CRC) and pancreatic (PC) cancers. Here, we implement the more comprehensive enhanced backscattering (EBS)spectroscopy to better understand the structural and optical changes which lead to the previous findings. EBS pro-vides high-resolution measurement of the spatial reflectance profile PðrsÞ between 30 microns and 2.7 mm, whereinformation about nanoscale mass density fluctuations in the mucosa can be quantified. A demonstration of thelength-scales at which PðrsÞ is optimally altered in CRC and PC field carcinogenesis is given and subsequently thesechanges are related to the tissue’s structural composition. Three main conclusions are made. First, the most sig-nificant changes in PðrsÞ occur at short length-scales corresponding to the superficial mucosal layer. Second, thesechanges are predominantly attributable to a reduction in the presence of subdiffractional structures. Third, similartrends are seen for both cancer types, suggesting a common progression of structural alterations in each.© TheAuthors.

Published by SPIE under a Creative Commons Attribution 3.0 Unported License. Distribution or reproduction of this work in whole or in part requires full

attribution of the original publication, including its DOI. [DOI: 10.1117/1.JBO.18.9.097002]

Keywords: enhanced backscattering; coherent backscattering; inverse scattering; optical properties; elastic light scattering; spectroscopy;colorectal cancer; pancreatic cancer.

Paper 130408R received Jun. 12, 2013; revised manuscript received Jul. 16, 2013; accepted for publication Aug. 7, 2013; publishedonline Sep. 5, 2013.

1 IntroductionCancer is a multistep process in which a number of smallermutations accumulate in a stepwise fashion that eventuallyleads to carcinoma and metastasis. One way to conceptualizethis process is through the notion of field carcinogenesis.1

Under this approach to understanding carcinogenesis, severalgenetic/epigenetic mutations are expressed diffusely throughouta diseased organ as subtle ultrastructural transformations thatconstitute a “fertile field” from which further cancer progressioncan arise. From within this field, localized neoplasia with malig-nant potential emerge through stochastic mutations.

Although the concept of field carcinogenesis has been understudy since 1953, the ultrastructural changes that compose thefield are still being uncovered. Several changes whose role infield carcinogenesis has been implicated include chromatincondensation,2–4 cytoskeleton disruption,5 and collagen cross-linking in the extracellular matrix.6 Still, the exact origin andcarcinogenic advantage these transformations confer are notfully understood.

The implications of more fully understanding the mecha-nisms behind field carcinogenesis are twofold. First, it providesa glimpse into cancer development at the earliest stages. Suchinformation can form a basis from which the later stages of

cancer can be better understood, and offers the potential foridentifying common origins behind all cancer types. Second,it can be exploited to accurately detect the presence of cancerat less invasive surrogate sites away from the location of neo-plasia. Moreover, detection of this kind could be performed at anearly stage where more treatment options are available and prog-nosis is vastly improved.

One technique that enables exploration into both implica-tions is optical characterization of biological tissue usingenhanced backscattering (EBS) spectroscopy, also known ascoherent backscattering.7 EBS is observed as an angular back-scattering peak whose shape is the two-dimensional (2-D)Fourier transform of the spatial reflectance profile Pð~rsÞ,where ~rs represents a relative spatial separation between theentrance and exit point of any multiply scattered ray in themedium. Previously, a variation on EBS known as low-coher-ence enhanced backscattering (LEBS) spectroscopy has demon-strated an ability to sense the structural alterations occurring infield carcinogenesis.8 LEBS uses partial coherence illuminationto selectively interrogate PðrsÞ at subdiffusion length-scales(rs < 1 transport mean free path). In a study of histologicallynormal rectal biopsies from 219 subjects undergoing colorectalcancer (CRC) screening, LEBS detected the presence ofadenomatous polyps >10 mm in diameter throughout thecolon with 100% sensitivity, 80% specificity, and 89.5% areaunder the receiver operator characteristic curve (AUROC).9 Inanother study of histologically normal duodenal biopsiesfrom 203 subjects undergoing upper endoscopy, LEBS detectedthe presence of pancreatic carcinomas with 95% sensitivity,

Address all correspondence to: Andrew J. Radosevich, Northwestern University,Department of Biomedical Engineering, Tech E310, 2145 Sheridan Road,Evanston, IL 60208. Tel: +847-467-3215; Fax: (847) 491-4928; E-mail: [email protected]

Journal of Biomedical Optics 097002-1 September 2013 • Vol. 18(9)

Journal of Biomedical Optics 18(9), 097002 (September 2013)


http://dx.doi.org/10.1117/1.JBO.18.9.097002






71% specificity, and 85% AUROC.10 These studies demon-strated the proof of principle that LEBS could detect the changesoccurring in human CRC and pancreatic cancer (PC) field car-cinogenesis. However, the LEBS instrumentation used in thesestudies was limited in two respects. First, it was not capable ofidentifying the structural and optical origin of the measuredchanges. Second, it could not be implemented for in vivoapplication.

To address these needs, two incarnations of the EBS instru-mentation are currently being developed. First, a bench-topinstrument provides comprehensive measurement of the shapeof PðrsÞ within both the subdiffusion and diffusion regimeswith high spatial resolution (∼4 μm).7,11 Measurements ofthis kind are extremely sensitive to the shape of the scatteringphase function,12 which contains information about the subtlestructural alterations occurring in field carcinogenesis at struc-tural length-scales down to ∼30 nm (Ref. 13). Second, a 3.4 mmdiameter probe-based LEBS instrument trades comprehensivemeasurement of PðrsÞ for the ability to fit inside the workingchannel of commercially available endoscopes.14 This configu-ration targets aspects of PðrsÞ that have demonstrated diagnosticpotential and has been presented for use in CRC and PC riskstratification.15–17

In this paper, we implement the comprehensive bench-topEBS instrument to study the structural origin behind the changespreviously observed in CRC and PC field carcinogenesis. Thepaper is organized as follows: In Sec. 2, we review the funda-mentals of scattering theory in random media under the Bornapproximation and the theoretical origin of the EBS phenome-non. Next, in Sec. 3, we detail the EBS instrument and dataprocessing, the use of EBS to extract ultrastructural and opticalproperties from experimental measurements, the experimentalprocedure used in our biopsy study, and the steps needed to gen-erate a 2-D random medium. In Sec. 4, we implement EBS todetermine the length-scales at which PðrsÞ is optimally altered,quantify the structural and optical alterations that lead to suchchanges, and estimate which depths contribute to the observedalterations. Finally, the conclusions are summarized in Sec. 5.

2 Theory

2.1 Scattering Theory: Born Approximation inContinuous Random Media

In this section, we briefly review the application of the Bornapproximation in a continuous random media to calculate

the scattering characteristics needed for the current study.Comprehensive discussion of scattering theory under theBorn approximation can be found in Chapter 13 of Ref. 18,Chapter 7 of Ref. 19, and Chapter 6 of Ref. 20. A more specifictreatment for continuous random media is found in Chapter 16of Ref. 21. Application to biological tissue using the Whittle-Matérn family of autocorrelation functions was first presentedin Refs. 22 and 23.

Scattering contrast within biological tissue is generatedby the random arrangement of a number of arbitrarily shapedstructures ranging in size from tens of nanometers to tens ofmicrons.24 The best way to mathematically describe such arandomly distributed media is through its statistical refractiveindex autocorrelation function BnðrdÞ. One versatile modelto describe BnðrdÞ uses the three-parameter Whittle-Matérnfamily of autocorrelation functions.22–25

BnðrdÞ ¼ An ·

�rdLn

�D−32

· KD−32

�rdLn

�; (1)

where the subscript on rd implies a differential separationbetween any two points, An is the amplitude of refractiveindex fluctuations [i.e., scales BnðrdÞ along y axis], Ln is thecharacteristic distribution length [i.e., scales BnðrdÞ alongx axis], and D determines the shape of the distribution.For rd < Ln, BnðrdÞ approaches a power law when0 < D < 3, a decaying exponential when D ¼ 4, and a Gaussianas D → ∞. Depictions of the spatial distribution of refractiveindex specified by BnðrdÞ for three values of D are shownin Fig. 1.

In this paper, we refer to An, Ln, and D as ultrastructuralproperties since they govern the spatial distribution of sub-diffractional fluctuations in refractive index. Additionally, werefer to values of rd as structural length-scales (e.g., subdiffrac-tional information is carried at short structural length-scalesbelow ∼λ∕2).

In order to relate BnðrdÞ to light scattering, we apply the firstBorn approximation to calculate the power spectral density Φsas the three-dimensional (3-D) Fourier transform of BnðrdÞ.21For the Whittle-Matérn model,Φs can be calculated analyticallyas22

ΦsðksÞ ¼AnL3

nΓ�D2

�π3∕22ð5−DÞ∕2 · ð1þ k2sL2

nÞ−D∕2; (2)

where ks ¼ 2k sin θ2and k is the wavenumber.

Fig. 1 Demonstration of the random media specified by the Whittle-Matérn model for (a) D ¼ 2, (b) D ¼ 3, and (c) D ¼ 4. The methods used togenerate these images are described in Sec. 3.3.


Radosevich et al.: Ultrastructural alterations in field carcinogenesis measured. . .


In order to describe the directionality of scattered light inten-sity, the differential scattering cross-section per unit volume forunpolarized light σðθÞ can be calculated by multiplying ΦsðksÞby the dipole radiation pattern.

σðθÞ ¼ 2πk4ð1þ cos2 θÞΦsð2k sinθ

2Þ: (3)

The shape of σðθÞ can be parametrized by the scatteringmean free path ls (i.e., inverse of the scattering coefficientμs) and the anisotropy factor g (i.e., how forward directedthe scattering is).20

ls ¼�2π

Z1

−1σðcos θÞd cos θ

�−1; (4)

g ¼ 2πls

Z1

−1cos θ · σðcos θÞd cos θ: (5)

Finally, the effective transport for a diffusely scatteringmedium is expressed by the transport mean free path.

l�s ¼ls

1 − g: (6)

We refer to ls, g, and l�s as optical properties.Analytical equations relating the ultrastructural and optical

properties under the Whittle-Matérn model can be found inRef. 22. One relationship that is useful to describe in furtherdetail is that between l�s and D. In the limit of kLn ≫ 1,

l�sðkÞ ∝ kD−4 for D < 4: (7)

In words, this means that for D < 4, the spectral dependenceof l�s follows a power-law distribution whose power is solelydependent on D. In this way, although D should primarily bethought of as an ultrastructural property, it is also conceptuallyuseful to think of it as an optical property that determines howstrongly light transport changes with illumination wavelength.

2.2 Enhanced Backscattering Theory

In this section, we briefly review the theoretical aspects of EBSthat are pertinent to the discussion in the following sections.Additional details concerning the origin of EBS can be found

in a number of seminal publications on the first experimentalobservation of EBS.26–29 More recently, publications from ourgroup detail the application of EBS to biological tissue charac-terization. These publications include an open source MonteCarlo program,30 detailed description of the bench-top instru-ment used to measure biological tissue,7 description of the meas-urement of the spatial reflectance profile using EBS,11

description of an inverse method to extract optical properties,31

and development of a miniaturized probe-based LEBS instru-ment to characterize tissue in vivo.14

The EBS signal is an angular intensity peak [IEBSðθx; θyÞ, seeFig. 2(a)] with its maximum value centered in the backscatteringdirection. The origin of IEBSðθx; θyÞ is the constructive interfer-ence between all time-reversed path-pairs within a scatteringmedium. Mathematically, the interference signal from allpath-pairs exiting the medium within the illumination spotcan be summed using a Fourier transform operation to obtainthe total observable signal.7

IEBSðθx; θyÞ ¼ZZ

∞

−∞Pðxs; ysÞ

· Sðxs; ysÞeikðxs sin θxþys sin θyÞdxsdys;

¼ FfPðxs; ysÞ · Sðxs; ysÞg; (8)

where the subscript on xs and ys implies a relative separationbetween any two points, Pðxs; ysÞ is the spatial distributionof all multiply scattered photons (i.e., two or more scattteringevents) exiting the medium in the exact backscattering direction(i.e., antiparallel to the incident beam), Sðxs; ysÞ modulates theshape due to finite spot-size illumination, and F represents theFourier transform operation.

Of primary interest for tissue characterization is functionPðxs; ysÞ, commonly known by a number of different namesincluding spatially resolved diffuse reflectance,32,33 spatial back-scattering impulse-response,11 diffuse reflectance profile, etc.While there are a great many instruments that measure someaspect of Pðxs; ysÞ,34–36 part of the power of using EBS isthat it provides noncontact measurement of Pðxs; ysÞ at veryshort length scales ranging from ðys; xsÞ ¼ �30 μm to�2.7 mm with ∼4 μm spatial resolution.11 This range of lengthscales enables comprehensive characterization of Pðxs; ysÞwithin both the subdiffusion (

ffiffiffiffiffiffiffiffiffiffiffiffiffiffixs þ ys

p< l�s) and diffusion

regimes, allowing for accurate extraction of all three ultrastruc-tural and optical properties.

θx (deg.)

(a)

θ y (de

g.)

IEBS

(θx,θ

y)

0.5 0 0.5

0.5

0

xs (µm)

(b)

y s (µm

)

log10

[P(xs,y

s)]

2000 0 2000

2000

0

0 1000

0

2

4

6

8x 10

−4

rs (µm)

(c)

P

P(rs)

Fig. 2 Example data showing average measurements from all colorectal cancer (CRC) control biopsy samples. (a) measured at λ ¼ 700 nm. (b) Log-scale Pðxs; ysÞ obtained through inverse Fourier transform. The red arrows indicate inaccuracies in the measurement of Pðxs; ysÞ due to noise ampli-fication as described in Sec. 3.1. (c) distribution obtained after integration over the azimuthal angle.




3 Materials and Methods

3.1 EBS Instrument and Data Processing

Detailed description and schematics for our EBS instrument canbe found in Refs. 7 and 13. In brief, broadband visible light froma supercontinuum laser source (SuperK Extreme EXW-6, NKTPhotonics A/S, Birkerød, Denmark) is coupled into a single-mode fiber (Thorlabs, Newton, New Jersey; mode field diameterof 4.6 μm at 680 nm) and collimated to an ∼10 mm beam diam-eter. The beam then passes through a linear polarizer and an iristhat reduces its diameter to 2 mm in size before being directedonto the biopsy. Light backscattered from the sample is thencollected by a 50∕50 nonpolarizing antireflection coated platebeam splitter, passes through a linear polarizer to detect thecopolarized channel, and is focused onto a CCD camera(PIXIS 1024B eXcelon, Princeton Instruments, Trenton, NewJersey) by a 100 mm focal length achromatic lens. A liquidcrystal tunable filter (VariSpec, PerkinElmer, Waltham,Massachusetts) attached to the camera separates the light intoits component wavelengths. This configuration detects angularbackscattering of up to 7.5776 deg with 0.0074 deg resolutionand wavelengths between 500 and 720 nm with a 20 nm filterbandwidth. After inverse Fourier transform, this provides aminimum spatial resolution of 3.78 μm (at λ ¼ 500 nm) anda maximum spatial range of 5.42 mm (at λ ¼ 700 nm).

To process the data, the biopsy measurement is first back-ground subtracted and then normalized by the total unpolarizedincoherent intensity measured from a reflectance standard (spec-tralon >98% reflectance, Ocean Optics, Dunedin, Florida).IEBSðθx; θyÞ is then obtained by subtracting the incoherent base-line with a plane fit, using data from an annular ring that is∼1.8 deg away from the peak maximum. An example of thepeak shape measured from a rectal biopsy with illuminationin the linear copolarized configuration is shown in Fig. 2(a).

The angular data are then converted to spatial coordinatesthrough inversion of Eq. (8) to obtain7,13,37

Pðxs; ysÞ ¼F−1IfEBSgðθx; θyÞ

Sðxs; ysÞ; (9)

where function S can be computed as the autocorrelation of thespatial illumination distribution incident on the biopsy normalizedto unity at the origin. Assuming the beam intensity is a top-hat function with diameter ∅, function S is found analytically as7

Sðxs;ysÞ

¼�

2∅2cos−1ðffiffiffiffiffiffiffiffiffix2sþy2s

p∕∅Þ−2

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiðx2sþy2s Þð∅2−x2s−y2s Þ

p∅2π

forffiffiffiffiffiffiffiffiffiffiffiffiffiffix2sþy2s

p<∅

0 forffiffiffiffiffiffiffiffiffiffiffiffiffiffix2sþy2s

p>∅

:

(10)

Figure 2(b) shows the Pðxs; ysÞ distribution extracted fromthe measurement of IEBSðθx; θyÞ in Fig. 2(a).

As a way to average noise and make it easier to displayexperimental results, we integrate Pðxs; ysÞ over the azimuthalangle to obtain PðrsÞ.

PðrsÞ ¼Z

2π

0

Pðxs ¼ rs cos ϕ; ys ¼ rs sin ϕÞdϕ: (11)

In words, PðrsÞ represents the total reflected intensity thatexits the medium in the backscattering direction within an

annulus of radius rs. Figure 2(c) shows the PðrsÞ distributioncalculated from Pðxs; ysÞ in Fig. 2(b). Matlab code performingthe operations in Eqs. (9) to (11) are posted on our laboratorywebsite.38

We note that since function S decays to 0 for rs > ∅, themeasurement of PðrsÞ is physically limited to length-scalessmaller than ∅ (2 mm in our case). Moreover, division bythe decaying function S increasingly amplifies noise at increas-ing rs, making the measurement inaccurate as rs → ∅ [indicatedby red arrows in Fig. 2(b)]. As such, in this paper, we only plotthe shape of PðrsÞ for length-scales in which function S is >5%.This occurs at ∼0.88∅, or, in our case 1760 μm.

3.2 EBS Extraction of Ultrastructural and OpticalProperties

Determination of the ultrastructural and optical properties ofeach biopsy sample is performed by fitting the shape of PðrsÞto the Whittle-Matérn model. This process was first presentedin Ref. 7. Briefly, a series of Monte Carlo simulations usingthe open source code detailed in Ref. 30 was performed inorder to generate a library of PðrsÞ for D ¼ 2.0 to 4.0 in 0.1steps, g ¼ 0.70 to 0.96 in 0.02 steps, and rs∕l�s ¼ 0.004 to 4in 0.004 steps. Through use of multidimensional interpolation,we are able to rapidly generate PðrsÞ for arbitrary optical pro-perties within the library bounds. In order to extract the threeultrastructural properties, we minimize the sum squared errorbetween the experimentally measured and library-generatedPðrsÞ · SðrsÞ using a gradient search method in Matlab. Wefit each curve over the values of rs for which we see a significantdifference between case and control. To avoid the contributionof absorption, we fit the data using the wavelength range from600 to 700 nm. Within this range, the contribution from absorp-tion is ∼1% that of scattering in the mucosa and is thereforenegligible. The three ultrastructural properties can then be con-verted to optical properties at each wavelength using the analyti-cal equations in Ref. 22.

It is important to make note of two assumptions/limitationsinherent in our fitting routine. First, the Monte Carlo simulationsare performed for the index matching case, and therefore assumethat there is no refractive index contrast at the boundary. In orderto account for this in our model, we use an empirically derivedscaling factor to adjust the magnitude of the Monte Carlo simu-lated PðrsÞ. Second, our model assumes a single slab geometry.As such, this technique is not capable of separating the con-tribution of layers with varying ultrastructural properties andinstead provides measurement of the bulk optical propertieswithin the slab.

3.3 Generation of Random Medium Realizations

To generate a single 2-D realization of the random media speci-fied by BnðrdÞ, the following steps can be followed: First, cal-culate the 2-D Fourier transform of BnðrdÞ and take the squareroot. This specifies the distribution of spatial frequencies presentin the medium. Second, multiply the resulting function by a 2-DGaussian white noise signal with 0 mean and a variance of 1.This creates randomness within the spatial frequency domain.Finally, return to spatial coordinates by calculating the inverse2-D Fourier transform and add the real and imaginary com-ponents. Microscopic resolution renderings can be made bymultiplying the spatial frequency domain function by an area




normalized top-hat function with radius k · NA, where NA is thedesired numerical aperture.

3.4 Experimental Biopsy Measurement Procedure

Two to three pinch biopsies (∼2 to 3 mm diameter) were takenfrom subjects undergoing normal endoscopic screening proce-dures in accordance with the institutional review board atNorthShore University Healthsystems. Biopsies <2 mm indiameter were not measured since they were smaller than theillumination beam diameter. For the CRC study, biopsieswere collected from histologically normal sites within the rectalmucosa of subjects undergoing colonoscopy. Progression offield carcinogenesis was categorized according to the size ofthe largest adenomatous polyp found during colonoscopyand verified by a hospital pathologist. Four groups wereassigned: healthy control (C, N ¼ 43) with no adenomatouspolyps found during colonoscopy, diminutive adenomas (DA,N ¼ 6) with diameter <5 mm, adenomas (A, N ¼ 25) withdiameter from 5 to 9 mm, and advanced adenomas (AA,N ¼ 19) with diameter >10 mm or adenocarcinoma. For thePC study, biopsies were collected from histologically normalperiampullary duodenum of subjects undergoing upper endos-copy procedures. Subjects were categorized into two groups:healthy controls (C, N ¼ 13) included individuals withoutany personal/family history of cancer or pancreato-biliary dis-ease and the PC group (N ¼ 19) included subjects harboringadenocarcinomas.

Biopsies from all subjects were transported to NorthwesternUniversity in phosphate buffered saline (PBS) solution andmeasured within 4 h of extraction. A minimum of five spectralEBS measurements (λ ¼ 500 to 700 nm in 20 nm steps) wereacquired from the epithelial side of each biopsy. To prepare thesample for measurement, it was placed in the center of a glassslide, oriented with the epithelium facing upward, and coveredwith PBS. Two 1-mm spacers were then placed on either side ofthe biopsy and then covered with a 165-μm glass cover-slip.This served to (1) ensure a uniform 1 mm thickness betweenall measurements, (2) maintain hydration for the experimentduration, and (3) direct the specular reflection away from thedetector.

After measurement with EBS, the biopsies were fixed in 10%formalin for a minimum of 24 h. The samples were then

embedded in paraffin wax and sectioned (5 μm) according tostandard protocols. Slides were stained with hematoxylin andeosin and imaged using a 10× microscope objective.

4 Results

4.1 PðrsÞ in Colorectal Cancer andPancreatic Cancer

Figure 3 shows the changes in PðrsÞ attributable to CRC fieldcarcinogenesis measured in ex vivo rectal biopsies. The generalshape of PðrsÞ in these biopsies is a strongly decaying functionof rs with ∼90% of the intensity falling within the first 1000 μmshown in Fig. 3(a). Going from the curve for control subjects (C)to those harboring advanced adenomas (AA), there is adownward and inward shift of PðrsÞ, indicated by the blackarrow. The center and thickness of each curve indicate themean and standard error, respectively. Performing a two-tailedstudent’s t-test at each rs, we find a statistical significance forvalues of rs ranging from ∼150 to 700 μm [shown as the shadedgray region in Fig. 3(a)].

In order to better visualize the location and magnitude ofthese changes, Fig. 3(b) shows the difference between PðrsÞfor control and AA subjects. For smaller values of rs, the effectincreases in magnitude, consistent with the argument that thealterations associated with field carcinogenesis are most pro-nounced in the superficial mucosal layers.4 As a way to deter-mine the rs location with the most significant changes, we findthe location in which the p-value is minimized. This value(rs;optimal) is found at 267 μm and indicated as the dotted circlein Fig. 3(b).

Finally, to demonstrate the magnitude of the effect duringthe early progression of CRC, Fig. 3(c) shows the normalizedintensity value of PðrsÞ at rs ¼ rs;optimal for the four subjectgroups ordered in increasing aggressiveness from left to right.Consistent with Figs. 3(a) and 3(b), we see a progression towarda lower intensity as the polyp aggressiveness increases with amaximal 18% decrease between control and AA subject groups.This effect is statistically significant at the 5% level for nondimin-utive adenomatous polyps (i.e., diameter > ¼ 5 mm), an obser-vation that is consistent with the earlier ex vivo CRC study.9

We note that the comparable curves for DA and A subjectsomitted from Figs. 3(a) and 3(b) follow the same progression as

0 1000

0

2

4

6

8x 10

−4

p<0.05

rs (µm)

(a)

P (

λ =

700

nm

)

AAC

0 1000−6

−4

−2

0

x 10−5

rs (µm)

(b)

PA

A −

PC

C DA A AA

0.8

0.9

1

1.1

1.2

(c)

I(r s =

rs,

optim

al)

(a.

u.)

p ~ 0.039

p ~ 0.002

N=43 N= 6 N=25 N=19

Rectal biopsy spatial reflectance profile

Fig. 3 CRC field carcinogenesis alterations measured in the shape of PðrsÞ from rectal biopsies. (a) Comparison between PðrsÞ for advanced adenomas(AA, diameter>10 mmor adenocarcinoma) versus control (C) with the thickness of the curve indicating the standard error betweenmeasurements. Thearrow indicates the direction of the change from AA to C and the gray shaded region indicates the values of rs for which the curves are significantlydifferent. (b) Difference (AA − C) between the curves shown in panel (a). The dotted circle indicates rs;optimal (i.e., the location of the most significanteffect) of the difference curve. (c) Normalized intensity at rs;optimal for groups at increasing risk of CRC. DA, diminutive adenomas with diameter from 0to 4 mm and A, adenomas with diameter from 5 to 9 mm.




demonstrated in Fig. 3(c). However, for figure clarity these datawere not shown.

We repeated the above analysis for the PC study. Figure 4shows the changes in PðrsÞ attributable to PC field carcinogen-esis measured in ex vivo duodenal biopsies. Going from thecurve for control subjects to those with PC, there is a downwardand inward shift of PðrsÞ, similar to the trend observed in CRC.We find a statistical significance over values of rs ranging from∼20 to 900 μm. Figure 4(b) shows the difference between PðrsÞfor control and PC subjects. As rs approaches 0, we finda monotonically increasing effect with rs;optimal ¼ 20 μm.Finally, Fig. 4(c) shows the normalized intensity value of

PðrsÞ at rs ¼ rs;optimal for control versus PC. This shows a56% reduction in intensity from control to PC, which isstatistically significant at the 5% level.

4.2 Ultrastructural Alterations in CRC and PC

In order to gain a more physical understanding of the structuralcomposition that leads to the observed alterations in PðrsÞ, weextracted the ultrastructural properties from the shape of PðrsÞas described in Sec. 3. We note that this extraction routineassumes that the medium is composed of a single slab of randommedia, which follows the Whittle-Matérn model. As such, the

0 10000

1

2

3

4x 10

−3

p<0.05

rs (µm)

(a)

P (

λ =

700

nm

)

PCC

0 1000

−2

−1

0x 10

−3

rs (µm)

(b)

PP

C −

PC

C PC0

0.5

1

(c)

I(r s =

rs,

optim

al)

(a.

u.) p ~ 0.004

N=13 N=19

Duodenal biopsy spatial reflectance profile

Fig. 4 Pancreatic cancer (PC) field carcinogenesis alterations measured in the shape of PðrsÞ from duodenal biopsies. (a) Comparison between PðrsÞ forPC versus control with the thickness of the curve indicating the standard error between measurements. The arrow indicates the direction of the changefrom PC to control and the gray shaded region indicates the values of rs for which the curves are significantly different. (b) Difference (PC − C) betweenthe curves shown in panel (a). The dotted circle indicates rs;optimal of the difference curve. (c) Normalized intensity at rs;optimal for PC and control groups.

0.01 0.1 1 100

2

4

6x 10

−5

rd (µm)

(a)

Bn

Rec

tal b

iop

sy u

ltra

−str

uct

ure

Diffraction lim

it λ/2

C AA

x (µm)

(b)

y (µ

m)

Control

1 2 3

2

1

0

x (µm)

(c)

AA

1.4 NA

Microscope

1 2 3

(b) (c)

−0.02 −0.01 0 0.01 0.02

x (µm)

(d)

y (µ

m)

Control

1 2 3

2

1

0

x (µm)

(e)

AA

CB

S sensitivity range

1 2 3

Fig. 5 Representations of the continuous randommedia extracted from the CRC study biopsy measurements. (a) BnðrdÞ for control versus AA. The grayshaded region indicates the approximate range of values over which enhanced backscattering (EBS) is sensitive. (b) and (c) Continuous media real-izations with 1.4 NA microscope resolution at λ ¼ 0.4 μm for the average ultrastructural properties of control and AA subjects, respectively. (d) and(e) Continuous media realizations over the EBS sensitivity range with the average ultrastructural properties of control and AA subjects, respectively.




extracted values provide estimates for the bulk ultrastructuraland optical properties over the entire biopsy.

Figure 5(a) shows the shape of BnðrdÞ corresponding to themean ultrastructural properties extracted for the control and AAsubjects in the CRC study. The gray shaded region indicatesthe approximate range of structural length-scales over whichEBS is sensitive to the shape of BnðrdÞ (∼30 nm to 3 μm).13

Interestingly, a reduction in the value of BnðrdÞ is seen withinthe subdiffractional regime (rd < λ∕2, λ ¼ 0.4 μm), but isnegligible for structural length-scales within the resolutionlimit of a conventional imaging system. This implies that thealterations observed in the shape of PðrsÞ are primarily attrib-utable to minuscule structures that could not be resolved in aconventional light microscope. To better convey this idea,Figs. 5(b) and 5(c) depict 1.4 NA microscope resolution render-ings of the random media described by the shapes of BnðrdÞ forcontrol and AA subjects, respectively. In these images, all sub-diffractional structures have been blurred away, leaving largerfluctuations that are qualitatively indistinguishable betweencontrol and AA subjects.

In contrast with the microscope resolution images, Figs. 5(d)and 5(e) show renderings of the comparable random media withresolution (30 nm) and spatial extent (3 μm) chosen such thatthey match the structural length scale sensitivity of EBS. In thisway, each rendering represents the range of structures that EBS“sees.” Comparison with the microscope resolution imagesdemonstrates the sensitivity of EBS to much finer compositionaldetails. Using this capability, it is possible to distinguishbetween control and AA subjects in Figs. 5(d) and 5(e), respec-tively. While the changes are very subtle, there is an appreciable

reduction in the presence of smaller structural length-scales forAA subjects.

Repeating this analysis for the PC study, Fig. 6(a) shows theshape of BnðrdÞ that corresponds to the mean ultrastructuralproperties extracted for all control and PC subjects. Similarto the CRC study above, a reduction in the value of BnðrdÞis seen within the subdiffractional regime, but becomesnegligible almost exactly at the diffraction limit. This higher-frequency detail is almost entirely blurred away in the micro-scope resolution images in Figs. 6(b) and 6(c). Depictionsof the range of structures detected by EBS for control andPC subjects are shown in Figs. 6(d) and 6(e), respectively. Inthis case, two prominent alterations are visible: (1) a reductionin refractive index variance and (2) a reduction in the presence ofsmaller structural length-scales.

4.3 Optical Property Alterations in CRC and PC

To provide quantification of the measured field carcinogenesischanges in a way that is more comparable to other optical tech-niques, we convert the ultrastructural properties into opticalproperties using the analytical equations presented in Ref. 22.The values of l�s measured at λ ¼ 700 nm, g measured atλ ¼ 700 nm, and D for CRC biopsies are shown in Fig. 7.We note that D is included as an optical property in this plotsince it demonstrates the spectral dependence of light transport(as described in Sec. 2). For all three optical properties, there isa statistically significant increase in value between C and AAsubjects, with l�s increasing 12.6%, g increasing 1.9%, andD increasing 5.1%.

0.01 0.1 1 100

1

2

3

4x 10

−4

rd (µm)

(a)

Bn

Du

od

enal

bio

psy

ult

ra−s

tru

ctu

re

Diffraction lim

it λ/2

C PC

x (µm)

(b)

y (µ

m)

Control

1 2 3

2

1

0

x (µm)

(c)

PC

1.4 NA

Microscope

1 2 3

( ) ( )

−0.05 0 0.05

x (µm)

(d)

y (µ

m)

Control

1 2 3

2

1

0

x (µm)

(e)

PC

CB

S sensitivity range

1 2 3

Fig. 6 Representations of the continuous random media extracted from the PC study biopsy measurements. (a) BnðrdÞ for control versus PC. The grayshaded region indicates the approximate range of values over which EBS is sensitive. (b) and (c) Continuous media realizations with 1.4 NAmicroscoperesolution at λ ¼ 0.4 μm for the average ultrastructural properties of control and PC subjects, respectively. (d) and (e) Continuous media realizationsover the EBS sensitivity range with the average ultrastructural properties of control and PC subjects, respectively.




Following the progression from C to A to AA, there is asmooth increase in value for each optical property. While theDA values in our dataset do not follow this trend, the large stan-dard error on their measurement (due to relatively small samplesize) does not preclude the possibility of a smooth progressionwith larger sample size. Indeed, based on the trend in Fig. 4(c)and the compounding multistep nature of carcinogenesis, this islikely the case.

The optical properties for biopsies in the PC study are shownin Fig. 8. Once again, for all three optical properties, there is astatistically significant increase in value, with l�s increasing21.1%, g increasing 4.6%, and D increasing 66.9%.Interestingly, the directionality of the optical property changesin field carcinogenesis is the same for both CRC and PC.

4.4 Estimate of the Penetration Depth of Changes inCRC and PC

To obtain a better understanding of the depths at which theobserved changes in PðrsÞ originate, we performed MonteCarlo simulations in which both rs and the maximum depthzmax of each photon were tracked to generate 2-D Pðrs; zmaxÞdistributions [shown in Figs. 9(a) and 10(a)]. Four simulationswith optical properties corresponding to the mean values calcu-lated for AA, CRC control, PC, and PC control were performed.For each simulation, a 2-D grid of Pðrs; zmaxÞ with fixed spatialcoordinates corresponding to the experimentally measuredrange of values was initiated. Light intensity was then depositedinto the fixed Pðrs; zmaxÞ grid positions until 1010 photon his-tories were computed.

C DA A AA3.2

3.4

3.6

3.8

(c)

D

p ~ 0.012

C DA A AA

0.83

0.84

0.85

0.86

(b)

g (λ

= 7

00 n

m)

p ~ 0.002

C DA A AA

1200

1300

1400

1500

(a)l s* (

µm;λ

= 7

00 n

m)

p ~ 0.007

Rectal biopsy optical properties

Fig. 7 Optical properties of rectal biopsies in the CRC study. (a) l�s measured at λ ¼ 700 nm. (b) g measured at λ ¼ 700 nm. (c) Each plot showsmeanþ standard error. N ¼ 43 for C, N ¼ 6 for DA, N ¼ 25 for A, and N ¼ 19 for AA.

C PC

2.2

2.4

2.6

2.8

3

(c)

D

p ~ 0.001

C PC

0.78

0.8

0.82

(b)

g ( λ

= 7

00 n

m)

p ~ 0.001

C PC

400

500

600

700

800

(a)

l s* (µm

;λ =

700

nm

)

p ~ 0.014

Duodenal biopsy optical properties

Fig. 8 Optical properties of duodenal biopsies in the PC study. (a) l�s measured at λ ¼ 700 nm. (b) g measured at λ ¼ 700 nm. (c) Each plot showsmeanþ standard error. N ¼ 13 for C and N ¼ 19 for PC.

log10

[PAA

]

rs (µm)

(a)

z max

(µm

)

500 1000 1500

200

400

600

800

−8

−7

−6

−5

rs (µm)

(b)

log10

[|PAA

−PC

|]

500 1000 1500

Rectal biopsy histology

200 µm

Crypts ~500 µm

(c)

Rectal biopsy penetration depth

Fig. 9 Penetration depth of in CRC field carcinogenesis using Monte Carlo simulation. (a) Log-scale Pðrs; zmaxÞ distribution for the average opticalproperties of all AA subjects. (b) Log-scale of the absolute difference in Pðrs; zmaxÞ between control and AA subjects. The dashed black line indicateszmaxðrsÞ. (c) Histological image from a rectal biopsy.




Figure 9 demonstrates the penetration depth of the PðrsÞ sig-nal in CRC field carcinogenesis. The log-scale PAAðrs; zmaxÞdistribution corresponding to the average optical propertiesof all AA subjects is shown in Fig. 9(a) (PC not shown sinceit is qualitatively very similar). As expected from multiplyscattered light transport, with increasing values of rs, thereis a corresponding increase in zmax. In order to isolate the loca-tion of the subtle differences in light propagation betweencontrol and AA subjects, Fig. 9(b) shows the absolute diffe-rence between PCðrs; zmaxÞ and PAAðrs; zmaxÞ in log-scale.Qualitatively, the location of the changes with the largest mag-nitude is found at very small values of rs and zmax. To betterquantify the depth of these changes, we calculate the averagepenetration depth of the difference signal for each value of rs as

zmaxðrsÞ ¼R 1000 μm0 zmax · ½PAAðrs; zmaxÞ−PCðrs; zmaxÞ�dzmaxR 1000 μm

0 ½PAAðrs; zmaxÞ−PCðrs; zmaxÞ�dzmax

:

(12)

The values of zmaxðrsÞ are represented as the dashed blackline in Fig. 9(b).

Conceptually, zmax represents the approximate depth atwhich the alterations occur (given the understanding that awide range of depths do, in fact, contribute). Comparing withthe values of rs for which the experimentally measuredCRC PðrsÞ data are significantly separated [data shown inFig. 3(a)], the depths at which the changes originate are between∼400 and 575 μm with the optimal difference occurring at465 μm in depth. To qualitatively visualize the morphologicalstructures interrogated over these depths, Fig. 9(c) shows a rep-resentative histological section acquired from one rectal biopsy.This image illustrates that the measured signal primarily origi-nates from within the colonic crypts (∼500 μm in length) andsurrounding extracellular matrix, which composes the superfi-cial mucosal layer.

Figure 10 repeats the penetration depth analysis of the PðrsÞsignal for PC field carcinogenesis. The log-scale PPCðrs; zmaxÞdistribution corresponding to the average optical properties ofall PC subjects is shown in Fig. 10(a). Figure 10(b) showsthe absolute difference between PC and PAA in log-scalewith the values of zmaxðrsÞ represented as the dashed blackline. Comparing with the values of rs for which the experimen-tally measured PC data are significantly separated [data shown

in Fig. 4(a)], the depths at which the signal originates arebetween ∼100 and 500 μmwith the optimal difference occurringat 100 μm in depth. The representative histological section fromone duodenal biopsy shown in Fig. 10(c) indicates that thesedepths correspond to the superficial mucosal layer, which iscomposed of villi, crypts, and the surrounding extracellularmatrix.

5 Discussion and ConclusionsIn many ways, EBS is the ideal technique to use when studyingbulk light transport in biological tissue.

From a utilization perspective, it offers an easy-to-implementinstrument that enables noncontact measurements of intrinsicscattering and absorption contrast (no staining needed). In addi-tion, due to the collimated beam delivery of light to the sample,EBS possesses a large tolerance in sample positioning (over20 cm range in sample positioning has no effect on the signal).This allows measurement of nonstationary targets to be acquiredwithout inducing error or motion artifacts. Moreover, since mea-surements are taken in reflection as opposed to transmission,tissue can be quantified in situ without sectioning or otherpreparation.

From a theoretical perspective, the high spatial resolutionmeasurement of PðrsÞ at subdiffusion length-scales is extremelysensitive to the shape of the differential scattering functionper unit volume σðθÞ.7,11,12 This allows extraction of not onlyparameters such as ls, g, and D, which define the shapeof σðθÞ, but also in principle the full shape of σðθÞ.Furthermore, the shape of σðθÞ is directly related to the shapeof BnðrdÞ through Fourier transformation, enabling quantifica-tion of ultrastructural components down to ∼30 nm in size (i.e.,approximately one order of magnitude smaller than the diffrac-tion limit).13 Finally, because the signal within the subdiffusionregime is primarily generated by the superficial depths of a scat-tering medium, an added benefit of EBS is its sensitivity to thesuperficial mucosal layers where epithelial cancers originate.Within the mucosal layer, specific depths can then be targetedby analyzing the signal at different values of rs.

37

In this paper, we utilized these benefits of EBS to determinethe values of rs for which PðrsÞ is significantly altered in fieldcarcinogenesis, extracted the structural composition and opticalproperties that lead to such changes, and estimated the depth atwhich the changes occur. Interestingly, in both the CRC and PC

log10

[PPC

]

rs (µm)

(a)

z max

(µm

)

500 1000 1500

200

400

600

800

−8

−7

−6

−5

−4

rs (µm)

(b)

log10

[|PPC

−PC

|]

500 1000 1500

Duodenal biopsy histology

200 µm

Villi ~ 300 µm

Crypts

(c)

Duodenal biopsy penetration depth

Fig. 10 Penetration depth of PðrsÞ in PC field carcinogenesis using Monte Carlo simulation. (a) Log-scale Pðrs; zmaxÞ distribution for the average opticalproperties of all PC subjects. (b) Log-scale of the absolute difference in Pðrs; zmaxÞ between control and PC subjects. The dashed black line indicateszmaxðrsÞ. (c) Histological image from a duodenal biopsy.




studies, we found a similar directionality of the changes inPðrsÞ, BnðrdÞ, and optical properties.

1. For PðrsÞ, with progression toward a more diseasedstate, we found a statistically significant downwardand inward shift in shape occurring at length-scalessmaller than ∼1 mm. The magnitude of the effect(quantified as the difference between disease and con-trol) was reduced with increasing values of rs, sug-gesting that structural alterations in the superficialmucosal layer are responsible for a majority of theobserved signal.

In CRC, we found a significant decrease in PðrsÞ forrs ¼ 150 to 700 μm, with the most significant changeoccurring at rs ¼ 267 μm. This corresponds to aver-age penetration depths between ∼400 and 575 μm,with the optimal difference occurring at 465 μm indepth. Morphologically, this signal is attributed tostructures within the colonic crypts and surroundingextracellular matrix.

In PC, we found a significant decrease in PðrsÞ forrs ¼ 20 to 900 μm, with the most significant changeoccurring at rs ¼ 20 μm. This corresponds to averagepenetration depths between ∼100 and 500 μm, withthe optimal difference occurring at 100 μm in depth.Morphologically, this signal is attributed to structureswithin the duodenal villi, crypts, and surroundingextracellular matrix.

2. For BnðrdÞ, there was an appreciable decrease in struc-tural correlation only for values of rd smaller than thediffraction limit of violet light (λ∕2 ¼ 200 nm). Thus,the physical alterations in field carcinogenesis mea-sured by EBS occur at structural length-scales smallerthan the resolution of a conventional light microscopeand, therefore, would be difficult or impossible todetect with such instruments. Physically, this changein BnðrdÞ is attributable to a reduction in the quantityof structures with size between 30 and 200 nm.

3. For each of the three optical properties (l�s , g, and D),we found an increase in value for the diseased tissue.In the CRC study, we found l�s increases 12.6%, gincreases 1.9%, and D increases 5.1%. In the PCstudy, we found an even larger effect with l�s increasing21.1%, g increasing 4.6%, and D increasing 66.9%.

Furthermore, the changes observed in the current study are inagreement with the decrease in LEBS enhancement factor E andspectral slope SS observed in the previous ex vivo bench-topLEBS PC and CRC studies9,10 and current in vivo opticalprobe LEBS PC and CRC studies.4,15–17 This can be confirmedby looking at the definitions of E and SS: The empirical param-eter E can be found as E ¼ ∫ ∞

0 PðrsÞ · CðrsÞdrs, where CðrsÞ isthe spatial coherence function [a decaying function that focuseson PðrsÞ at short length scales]. As such, the decrease in PðrsÞ atshort length scales observed in the current study is consistentwith a decrease in E. Moreover, this decrease in E is consistentwith the increase in l�s in the current study since E ∝ 1∕l�s(Ref. 31). The empirical parameter SS is the negative slopeof a linear regression line fit to EðλÞ and can be related to D

as D ¼ −SS · hλihEi−1 þ 3, where h·i indicates the averageover the entire wavelength range.31 Thus, the increase in Dfor the current study is consistent with a decrease in SS. Theconsistency between all studies provides corroboration for thechanges presented in the current study.

The fact that the observed changes occurred with similardirectionality for both the CRC and PC studies suggests theexistence of a common progression of structural transformationsin the early stages of these two types of cancer (and potentiallyfor a majority or even all cancers). In other words, although bothcancers emerge through a unique sequence of genetic altera-tions, the resulting structural phenotype manifests itself in asimilar way as both diseases progress. This is understandableas the fundamental objective of all cancers is to create an envi-ronment that promotes cell survival, proliferation, and eventualmetastasis.

While the statistical nature of these ultrastructural alterationswas quantified in the current study, the specific structures thatcontribute to the observed effect cannot be determined directlyusing EBS. Nonetheless, we can conclude that the signal islocalized to superficial depths within the mucosa. The implica-tion for clinical diagnosis is that short penetration depths needto be targeted in order to achieve optimal performance.Mechanistically, this provides a general road map of which mor-phological structures to investigate. Ongoing research to impli-cate specific mucosal components suggests that the alterationsoccur due to a combination of intracellular (e.g., condensation ofchromatin in the nucleus2–4 or abnormalities in the cytoskele-ton5) and extracelluar components (e.g., collagen cross-link-ing4,6). Interestingly, these changes appear to be attenuatedversions of other classical markers of neoplasia.4 Chromatincompaction at larger length-scales (>0.5 μm) is a hallmark ofdysplasia widely used in histopathology,39 and lysyl oxidase(LOX)-induced collagen cross-linking is a well-known hallmarkof tumor microenvironment.40,41 Thus, the changes detected infield carcinogenesis are largely the same as those occurring atthe site of a focal tumor, only muted in magnitude. A forthcom-ing paper will provide 3-D spatial maps of the ultrastructural/optical alterations occurring within CRC and PC mucosausing inverse spectroscopic optical coherence tomography.42–44

AcknowledgmentsThis study was supported by National Institutes of Health grantnumbers RO1CA128641 and R01EB003682. A. J. Radosevichis supported by a National Science Foundation GraduateResearch Fellowship under Grant No. DGE-0824162.

References1. D. P. Slaughter, H. W. Southwick, and W. Smejkal, “Field cancerization

in oral stratified squamous epithelium. Clinical implications of multi-centric origin,” Cancer 6(5), 963–968 (1953).

2. Y. Stypula-Cyrus et al., “Hdac up-regulation in early colon field carcino-genesis is involved in cell tumorigenicity through regulation of chroma-tin structure,” PLoS ONE 8(5), e64600 (2013).

3. J. S. Kim et al., “The influence of chromosome density variations on theincrease in nuclear disorder strength in carcinogenesis,” PhysicalBiology 8(1), 015004 (2011).

4. V. Backman and H. K. Roy, “Advances in biophotonics detection offield carcinogenesis for colon cancer risk stratification,” J. Cancer4(3), 251–261 (2013).

5. N. N. Mutyal et al., “Biological mechanisms underlying structuralchanges induced by colorectal field carcinogenesis measured with




http://dx.doi.org/10.1002/(ISSN)1097-0142

http://dx.doi.org/10.1371/journal.pone.0064600

http://dx.doi.org/10.1088/1478-3975/8/1/015004

http://dx.doi.org/10.1088/1478-3975/8/1/015004

http://dx.doi.org/10.7150/jca.5838

low-coherence enhanced backscattering (LEBS) spectroscopy,” PLoSOne 8(2), e57206 (2013).

6. V. Backman and H. K. Roy, “Light-scattering technologies for fieldcarcinogenesis detection: a modality for endoscopic prescreening,”Gastroenterology 140(1), 35–41 (2011).

7. A. J. Radosevich et al., “Polarized enhanced backscattering spec-troscopy for characterization of biological tissues at subdiffusionlength scales,” IEEE J. Sel. Topics Quantum Electron. 18(4), 1313–1325 (2012).

8. Y. L. Kim et al., “Depth-resolved low-coherence enhanced backscatter-ing,” Opt. Lett. 30(7), 741–743 (2005).

9. H. K. Roy et al., “Association between rectal optical signatures andcolonic neoplasia: potential applications for screening,” Cancer Res.69(10), 4476–4483 (2009).

10. V. Turzhitsky et al., “Investigating population risk factors of pancreaticcancer by evaluation of optical markers in the duodenal mucosa,” Dis.Markers 25(6), 313–321 (2008).

11. A. J. Radosevich et al., “Measurement of the spatial backscatteringimpulse-response at short length scales with polarized enhanced back-scattering,” Opt. Lett. 36(24), 4737–4739 (2011).

12. E. Vitkin et al., “Photon diffusion near the point-of-entry in anisotropi-cally scattering turbid media,” Nat. Commun. 2, 587 (2011).

13. A. J. Radosevich et al., “Structural length-scale sensitivities of reflec-tance measurements in continuous random media under the bornapproximation,” Opt. Lett. 37(24), 5220–5222 (2012).

14. N. N. Mutyal et al., “A fiber optic probe design to measure depth-limitedoptical properties in-vivo with low-coherence enhanced backscattering(LEBS) spectroscopy,” Opt. Express 20(18), 19643–19657 (2012).

15. N. N. Mutyal et al., “In-vivo risk stratification of pancreatic cancerby evaluating optical properties in duodenal mucosa,” in BiomedicalOptics and 3-D Imaging, p. BSu5A.10, Optical Society of America,Washington, DC (2012).

16. N. N. Mutyal et al., “In vivo pancreatic cancer risk analysis by duodenalinterrogation with low coherence enhanced backscattering spectros-copy,” Clin. Can. Res. (in preparation).

17. N. N. Mutyal et al., “In-vivo risk stratification of colon carcinogenesisby measurement of optical properties with novel lens-free fiber opticprobe using low-coherence enhanced backscattering spectroscopy(LEBS),” in Conference 8230: Biomedical Applications of LightScattering VI (2012).

18. M. Born and E. Wolf, Principles of Optics: Electromagnetic Theory ofPropagation, Interference and Diffraction of Light, 7th ed., CambridgeUniversity Press, Cambridge, UK (1999).

19. H. van de Hulst, Light Scattering by Small Particles, Dover, New York(1981).

20. C. F. Bohren and D. R. Huffman, Absorption and Scattering of Light bySmall Particles, John Wiley and Sons, New York (1983).

21. A. Ishimaru, Wave Propagation and Scattering in Random Media.Volume I: Single Scattering and Transport Theory, Academic Press,New York (1978).

22. J. D. Rogers, İ. R. Çapoğlu, and V. Backman, “Nonscalar elastic lightscattering from continuous random media in the born approximation,”Opt. Lett. 34(12), 1891–1893 (2009).

23. C. J. R. Sheppard, “Fractal model of light scattering in biological tissueand cells,” Opt. Lett. 32(2), 142–144 (2007).

24. J. M. Schmitt and G. Kumar, “Turbulent nature of refractive-index var-iations in biological tissue,” Opt. Lett. 21(16), 1310–1312 (1996).

25. P. Guttorp and T. Gneiting, “Studies in the history of probability andstatistics xlix on the matérn correlation family,” Biometrika 93(4),989–995 (2006).

26. Y. Kuga and A. Ishimaru, “Retroreflectance from a dense distributionof spherical particles,” J. Opt. Soc. Am. A 1(8), 831–835 (1984).

27. L. Tsang and A. Ishimaru, “Backscattering enhancement of randomdiscrete scatterers,” J. Opt. Soc. Am. A 1(8), 836–839 (1984).

28. P. E. Wolf and G. Maret, “Weak localization and coherent backscatter-ing of photons in disordered media,” Phys. Rev. Lett. 55(24), 2696–2699(1985).

29. M. P. V. Albada and A. Lagendijk, “Observation of weak localization oflight in a random medium,” Phys. Rev. Lett. 55(24), 2692–2695 (1985).

30. A. J. Radosevich et al., “Open source software for electric field MonteCarlo simulation of coherent backscattering in biological media contain-ing birefringence,” J. Biomed. Opt. 17(11), 115001 (2012).

31. V. Turzhitsky et al., “Measurement of optical scattering properties withlow-coherence enhanced backscattering spectroscopy,” J. Biomed. Opt.16(6), 067007 (2011).

32. T. J. Farrell, M. S. Patterson, and B. Wilson, “A diffusion theory modelof spatially resolved, steady-state diffuse reflectance for the noninvasivedetermination of tissue optical properties in vivo,” Med. Phys. 19(4),879–888 (1992).

33. A. Kienle et al., “Spatially resolved absolute diffuse reflectancemeasurements for noninvasive determination of the optical scatteringand absorption coefficients of biological tissue,” Appl. Opt. 35(13),2304–2314 (1996).

34. R. Reif, O. A’Amar, and I. J. Bigio, “Analytical model of light reflec-tance for extraction of the optical properties in small volumes of turbidmedia,” Appl. Opt. 46(29), 7317–7328 (2007).

35. A. Amelink and H. J. C. M. Sterenborg, “Measurement of the local opti-cal properties of turbid media by differential path-length spectroscopy,”Appl. Opt. 43(15), 3048–3054 (2004).

36. G. Zonios et al., “Diffuse reflectance spectroscopy of human adenoma-tous colon polyps in vivo,” Appl. Opt. 38(31), 6628–6637 (1999).

37. A. J. Radosevich et al., “Depth-resolved measurement of mucosalmicrovascular blood content using low-coherence enhanced backscat-tering spectroscopy,” Biomed. Opt. Express 1(4), 1196–1208 (2010).

38. A. J. Radosevich and J. D. Rogers, “Biophotonics Laboratory atNorthwestern University,” http://biophotonics.bme.northwestern.edu/resources/index.html (25 August 2013).

39. D. Zink, A. H. Fischer, and J. A. Nickerson, “Nuclear structure in cancercells,” Nat. Rev. Cancer 4(9), 677–687 (2004).

40. H. E. Barker, T. R. Cox, and J. T. Erler, “The rationale for targeting theLOX family in cancer,” Nat. Rev. Cancer 12(8), 540–552 (2012).

41. A.-M. Baker et al., “The role of lysyl oxidase in src-dependent prolif-eration and metastasis of colorectal cancer,” J. Natl. Cancer Inst. 103(5),407–424 (2011).

42. J. Yi et al., “Spatially-resolved optical and ultrastructural properties inthe field effect of pancreatic and colorectal cancer observed by inversespectroscopic optical coherence tomography,” J. Biomed. Opt. (underreview).

43. J. Yi et al., “Can OCT be sensitive to nanoscale structural alterations inbiological tissue?,” Opt. Express 21(7), 9043–9059 (2013).

44. J. Yi and V. Backman, “Imaging a full set of optical scattering propertiesof biological tissue by inverse spectroscopic optical coherence tomog-raphy,” Opt. Lett. 37(21), 4443–4445 (2012).






http://dx.doi.org/10.1053/j.gastro.2010.11.023

http://dx.doi.org/10.1109/JSTQE.2011.2173659

http://dx.doi.org/10.1364/OL.30.000741

http://dx.doi.org/10.1158/0008-5472.CAN-08-4780

http://dx.doi.org/10.1155/2008/958214

http://dx.doi.org/10.1155/2008/958214

http://dx.doi.org/10.1364/OL.36.004737

http://dx.doi.org/10.1038/ncomms1599

http://dx.doi.org/10.1364/OL.37.005220

http://dx.doi.org/10.1364/OE.20.019643

http://dx.doi.org/10.1364/OL.34.001891

http://dx.doi.org/10.1364/OL.32.000142

http://dx.doi.org/10.1364/OL.21.001310

http://dx.doi.org/10.1093/biomet/93.4.989

http://dx.doi.org/10.1364/JOSAA.1.000831

http://dx.doi.org/10.1364/JOSAA.1.000836

http://dx.doi.org/10.1103/PhysRevLett.55.2696

http://dx.doi.org/10.1103/PhysRevLett.55.2692


http://dx.doi.org/10.1117/1.3589349

http://dx.doi.org/10.1118/1.596777

http://dx.doi.org/10.1364/AO.35.002304

http://dx.doi.org/10.1364/AO.46.007317

http://dx.doi.org/10.1364/AO.43.003048

http://dx.doi.org/10.1364/AO.38.006628

http://dx.doi.org/10.1364/BOE.1.001196

http://biophotonics.bme.northwestern.edu/resources/index.html






http://dx.doi.org/10.1038/nrc1430

http://dx.doi.org/10.1038/nrc3319

http://dx.doi.org/10.1093/jnci/djq569

http://dx.doi.org/10.1364/OE.21.009043

http://dx.doi.org/10.1364/OL.37.004443

Ultrastructural alterations in field carcinogenesis measured by ...€¦ · Ultrastructural...

Documents

Transcript of Ultrastructural alterations in field carcinogenesis measured by ...€¦ · Ultrastructural...