13 June 2017 Detection of precancerous lesions in the oral cavity using oblique polarized reflectance spectroscopy: a clinical feasibility study
Author Affiliations +
J. of Biomedical Optics, 22(6), 065002 (2017). doi:10.1117/1.JBO.22.6.065002
Abstract
We developed a multifiber optical probe for oblique polarized reflectance spectroscopy (OPRS) in vivo and evaluated its performance in detection of dysplasia in the oral cavity. The probe design allows the implementation of a number of methods to enable depth resolved spectroscopic measurements including polarization gating, source–detector separation, and differential spectroscopy; this combination was evaluated in carrying out binary classification tasks between four major diagnostic categories: normal, benign, mild dysplasia (MD), and severe dysplasia (SD). Multifiber OPRS showed excellent performance in the discrimination of normal from benign, MD, SD, and MD plus SD yielding sensitivity/specificity values of 100%/93%, 96%/95%, 100%/98%, and 100%/100%, respectively. The classification of benign versus dysplastic lesions was more challenging with sensitivity and specificity values of 80%/93%, 71%/93%, and 74%/80% in discriminating benign from SD, MD, and SD plus MD categories, respectively; this challenge is most likely associated with a strong and highly variable scattering from a keratin layer that was found in these sites. Classification based on multiple fibers was significantly better than that based on any single detection pair for tasks dealing with benign versus dysplastic sites. This result indicates that the multifiber probe can perform better in the detection of dysplasia in keratinized tissues.
Bailey, Verma, Fradkin, Lam, MacAulay, Poh, Markey, and Sokolov: Detection of precancerous lesions in the oral cavity using oblique polarized reflectance spectroscopy: a clinical feasibility study

1.

Introduction

The early detection of oral cavity cancer can greatly reduce morbidity rates as the 5-year survival rate associated with localized stage increases to 83% from 36% for the disease that has a distant spread.1,2 Even in highly developed nations where dental exams are prevalent, most cases of oral cancer are not detected until large, symptomatic lesions exist and the disease has advanced beyond the organ site when treatment options are limited and less effective.1 Suspicious lesions are typically biopsied following visual inspection and physical palpitation. Unfortunately, current screening techniques are limited by various confounding factors that can mask oral cancer progression. For example, benign inflammatory conditions appear very similar to premalignant and malignant lesions, making it difficult for even highly trained physicians to differentiate them. Additionally, one of the biggest risk factors in oral cancer is the synergistic effect of alcohol and tobacco, which exposes the entire lining of the oral cavity causing some of the lesions to span large regions of the cavity.3 Since biopsies can be taken only from a few suspicious regions, there is a high probability of sampling errors in oral cancer detection and diagnosis. Furthermore, after treatment, oral cancer patients require monitoring for potential development of secondary tumors that involves routine biopsies of their oral cavity taken over the course of many years—a painful invasive procedure that is also prone to sampling errors.4 Therefore, there is an evident need for a noninvasive method that would provide a real-time feedback to facilitate directed biopsies and to aid in the detection and monitoring of premalignant lesions in the oral cavity.

Optical techniques have emerged as promising tools in addressing challenges associated with the detection of oral cancer.56.7.8 Fluorescence imaging and spectroscopy have been used to detect changes in tissue fluorescence associated with endogenous fluorophores, such as reduced nicotinamide adenine dinucleotide (NADH) and flavin adenine dinucleotide (FAD), in the epithelium and collagen/elastin in the stroma.910.11.12.13 It was shown that the fluorescence of the epithelial cells increases and the fluorescence of the stroma decreases with neoplasia.14 The net result of these changes in situ is an overall decrease in fluorescence intensity with blue light excitation.15 This property has led to the development of VELscope (LED Dental Inc., White Rock, British Columbia)—a commercial device that is used as an adjuvant device in the visualization of oral cavity precancerous and cancerous lesions. Although very useful, this device has a relatively low specificity due to false positive loss of fluorescence in the areas of inflammation.16,17 Therefore, there is an ongoing effort to improve specificity in the examination of the oral cavity. Optical coherence tomography (OCT) has been extensively studied in imaging of oral tissue for the evaluation of oral carcinogenesis. OCT can provide high-resolution images of the entire thickness of the oral epithelium but requires interpretation by a trained histopathologist in assessment of live images.6,18 Raman spectroscopy has been investigated as a diagnostic tool to discriminate cancerous from normal oral tissue by characterizing chemical and molecular tissue composition.19,20 Although Raman spectroscopy can be very informative, Raman signals are very weak, making implementation of this technology quite challenging. Elastic scattering spectroscopy (ESS) has also shown potential in detecting quantitative morphological and structural information of the oral tissue.2122.23 However, significant challenges are associated with ESS interpretation, including a low signal-to-noise ratio.

There are multiple factors that complicate sensitive detection of malignancies in the oral cavity, including high levels of keratinzation and variable epithelial thickness due to benign inflammatory conditions. Furthermore, neoplastic changes of epithelia and the underlying stroma in the oral cavity are associated with concurrent alterations in optical signals from different depths.14,15,24,25 These complications require the development of depth-sensitive approaches that would allow simultaneous evaluation of optical signals from various depths in tissue. However, acquiring depth-dependent alterations in optical signatures associated with an oral malignancy and, especially, separating signals from the epithelium and stroma is a challenging task. Indeed, the avascular epithelium is thin and optically transparent, whereas the stroma underneath contains a dense network of collagen and elastin fibers, various cells, and blood vessels causing a strong scattering and hemoglobin absorption that dominates optical signals. Several methods have been implemented to improve signal collection from the superficial epithelial layer using depth-sensitive spectroscopy; these approaches can provide quantitative morphological and architectural information associated with oral precancer from targeted depths in the epithelium and stroma and, thus, can become useful clinical tools in the detection and monitoring of oral cancer. To this end, a number of probe designs with optical fibers normal to the tissue surface were evaluated with depth selectivity achieved by applying a variable aperture,26,27 variable source–detector separations,2728.29 and differential path length spectroscopy.30 However, these designs were limited in their ability to isolate the relatively weak optical signatures of the epithelium from the dominating stromal signal. An improvement in the detection of photons originating in the superficial epithelial layer was demonstrated using optical probes with a spherical lens10,25,31 or by positioning fibers in an oblique orientation.3233.34.35.36 In addition to changing probe geometry, our group and others developed polarization gating to separate epithelial signals from the diffuse background of the underlying stroma.3738.39.40.41.42.43 This approach is based on a combination of linear polarized illumination and collection of scattered light with polarization parallel and perpendicular relative to the illumination. Detected photons that travel a short distance within a sample undergo a single or a small number of scattering events, thus, maintaining their original polarization state. Conversely, photons that propagate deeper undergo many scattering events and, as a result, have random polarization. Thus, detection of photons that preserve their original polarization state provides a method to isolate the epithelial scattering. Encouraging clinical results have demonstrated the potential of polarization gating spectroscopy in the detection of colonic carcinogenesis and pancreatic adenocarcinoma.2,44

To further improve depth selectivity in reflectance spectroscopy, our group combined an oblique collection fiber geometry with polarization gating in a method termed oblique polarized reflectance spectroscopy (OPRS).24,32 Evaluation of this method in a pilot clinical trial in the oral cavity showed 90% sensitivity and 86% specificity in the separation of normal tissue from high-grade dysplasia and carcinoma. After these promising pilot clinical studies, we hypothesized that the performance of the probe can be significantly improved if a new probe design accounts for variations in thicknesses of the epithelium and the keratin layers that are very common in patients with malignancies in the oral cavity. To test this hypothesis, we designed a compact OPRS probe for simultaneous collection of polarized reflectance spectroscopic signals from multiple depths using multiple angle-polished beveled detector fibers (BF).45 The probe’s simple design uses a small number of components and no moving parts, which simplifies manufacturing and has low production costs. Encouraging results obtained with multilayer, tissue-mimicking phantoms, and in vivo measurements of normal oral mucosa demonstrated feasibility of the multifiber OPRS probe to provide depth-resolved measurements within tissue.45 Here, we evaluated the ability of this multifiber OPRS technology in the detection of dysplastic changes in patients with malignancies in the oral cavity.

2.

Methods

2.1.

Experimental System

A schematic of the OPRS system is shown in Fig. 1(a). An illumination fiber (100-μm core diameter, NA=0.12) delivered light to tissue from a 20-W tungsten halogen broadband light source (Ocean Optics, HL2000HP-FHSA). Then, light scattered from the tissue was collected by detection fibers (100-μm core diameter, NA=0.12) and was delivered to an imaging spectrometer (PI Acton SpectraPRO SP-2356, Pixis 2KB) equipped with a 150  g/mm grating optimized for the visible wavelength region (500-nm blaze). To enable a modular design wherein multiple probes could be easily interchanged, proximal ends of illumination and detection fibers were connected using low insertion loss (ca. 0.15 dB) FC/APC adapters to coupling fibers leading to either the light source (the illumination fiber) or the imaging spectrometer (detection fibers). At the spectrometer, detection fibers were assembled in a vertical array using a slit ferrule for alignment with the spectrometer’s entrance slit. Light from all collection fibers was simultaneously dispersed by the spectrometer’s grating onto an imaging CCD. An image was produced with the vertical dimension corresponding to different detection fibers and the horizontal axis displaying scattering spectra collected by the fibers thus simultaneously capturing spectra from all detection fibers in a single image. The entire system was housed on a wheeled cart for mobility [Fig. 1(b)].

Fig. 1

(a) Schematic of the overall system for OPRS. (b) Picture of the clinical system on a wheeled cart. (c) Illustration of the distal end of the OPRS probe: I denotes the illumination fiber; BF labels individual detection fibers where BF1, BF2, and BF3 collected copolarized (parallel) component of tissue scattering and BF1per, BF2per, and BF3per collected cross-polarized (perpendicular) scattering. (d) Artistic three-dimensional rendering and (e) an actual image of the distal end of OPRS after fiber polishing (scale bar: 1 mm). (f) Orientation of polarization transmission axes of polarizing film relative to the fiber array of the OPRS probe.

JBO_22_6_065002_f001.png

A schematic of the distal end of the probe for OPRS is shown in Fig. 1(c). All fibers had a silica core with n=1.458 and diameter of 100  μm; fluorine-doped cladding with n=1.455 and diameter—110  μm; and an NA of 0.12 (CeramOptec Industries, Inc., WF 100/110 P12). Obliquely oriented collection fibers are referenced according to their distance from the illumination fiber—BF1 is separated from the illumination fiber by one flat spacer fiber; BF2 is separated from the illumination fiber by two fibers, etc. Two flat tip fibers were positioned on each side of the illumination fiber as spacers to accommodate the gap between two pieces of the polarizing film that was used for polarized light illumination/detection as described below. The detection fibers (BFs) were polished at 40 deg with respect to the sample surface using a custom-made polishing puck. We have reported previously that this angle provides the optimum combination of depth selectivity and collection efficiency.45 In this multifiber design, collection cones of detection fibers and the illumination beam overlap at progressively increasing depth in tissue as the distance between the illumination and detection fibers increases thus allowing control over the sampling depth of tissue optical properties. The diameter of the fiber array of the constructed probe was ca. 1.6 mm. The outer probe diameter was made 7.6 mm for handling convenience in applications for detection of precancers in the oral cavity. An artistic rendering and an image of the distal end of the OPRS probe is shown in Figs. 1(d) and 1(e), respectively, to illustrate the beveled probe design following angle polishing. Once the fibers were polished, two pieces of 150-μm thick polarizing film (n=1.458) with 0.0002% extinction transmittance of cross-polarized light were glued to the distal end of the fibers using optically transparent and biocompatible epoxy (Epo-Tek 301-2). The two polarizing film pieces were positioned with their polarization axis perpendicular to each other as shown in Fig. 1(f), which allowed one half of the detection fibers to collect light copolarized with the illumination light polarization and the other half to collect cross-polarized light. Here, the scattered light collected with polarization parallel to the illumination is termed parallel and the cross-polarized light is defined as perpendicular. The fiber assembly with polarizing films was secured inside steel tubing with the outer diameter of 7.6 mm using biocompatible epoxy glue (Epo-Tek 301-2). Then, the steel tubing was carefully bent at an angle of 45 deg for easy positioning inside the oral cavity. Finally, a 200-μm thick quartz window (n=1.54) was glued to the polarizing film for protection of the probe during sterilization and to achieve an optimum overlap between illumination and collection cones at tissue surface.

2.2.

Collection of Clinical Data

In vivo spectra were collected with informed consent from 28 patients who were 18 years old and over and who were referred to the Department of Integrative Oncology at the British Columbia Cancer Agency (BCCA) with lesions in the oral cavity suspicious for dysplasia or carcinoma. The spectroscopic measurements followed standard oral cavity examinations by a physician. The spectra were collected from all sites suspicious for dysplasia and from contralateral (whenever possible) normal sites. The abnormal sites were biopsied and were histologically confirmed as benign, mild dysplasia (MD), or severe dysplasia (SD). Benign sites appeared abnormal during clinical examination but were histologically diagnosed as normal. Several (2 to 3) normal site measurements were taken for each patient. Depending on the size of an abnormal lesion, one to several measurements per lesion were obtained. All measurements were taken with room lights turned off to minimize background signal. Calibration spectra were acquired before each patient evaluation using a diffuse reflectance substrate standard (SRS-99, Labsphere, Inc.) and a background signal was measured using minimally reflecting black substrate (SRS-02, Labsphere, Inc.). Measurements from three patients were removed due to errors during data collection process associated with user mishandling of the probe or system malfunctioning.

2.3.

Preprocessing of Oblique Polarized Reflectance Spectroscopy Spectra

The mathematical equation that was used in spectra preprocessing is given below and it includes the following steps. First, the background signal from the minimally reflecting black substrate (Idark) was subtracted from the collected raw spectra (Imeas) to remove residual environmental light. Background signal from the environment was minimized by carrying out measurements in a dark room. Next, the spectral responses of the source, fibers, and detector were accounted for by normalizing the background corrected spectra by the spectrum from the diffuse reflectance substrate standard (Iwhte). Because the diffuse reflectance standard is not a perfect depolarizer,46 the perpendicular signal of the reflectance standard was multiplied by a ratio of the parallel to the perpendicular component called the depolarization coefficient (D) to account for this effect. To minimize background from residual room light, background signals (Idark) were first subtracted from the parallel and perpendicular spectra. In addition, collection areas (A) at the tissue interface for each detection fiber was determined in Zemax (Zemax, LLC, Kirkland, Washington) and was used to correct for the trend that fibers farther from the illumination fiber collect scattered photons from larger areas. To correct for the variations in the collection efficiency of detection fibers, the spectra were divided by the power throughput of each fiber (P). The relative power throughputs of detection fibers (Ppar and Pper) were determined by connecting each collection fiber’s promixal end to the illumination light source and measuring the power at the fiber’s distal end with a power meter (371R Optical Power Meter, Graseby Optronics). Also, differences in the collection time (t) that was used during data acquisition were accounted for. In summary, the normalization scheme to produce comparable OPRS data was as

I(λ)=Imeas,(λ)Idark,(λ)Iwhite,(λ)Idark,(λ)·1t·1P·1A,I(λ)=Imeas,(λ)Idark,(λ)[Iwhite,(λ)Idark,(λ)]·D·1t·1P·1A,D=Iwhite  Idark,IwhiteIdark,,
where I(λ) and I(λ) are the normalized parallel and perpendicular spectra, respectively.

2.4.

Quantifying Penetration Depth of Collection Fibers

To quantify the depth in a turbid media from which collection fibers collect scattering signal, an experiment was adapted from a method developed in Refs. 38 and 47. Briefly, a glass container with a 2-cm thick layer of optically transparent cured polydimethylsiloxane (PDMS) at the bottom to eliminate any back-reflections was filled with 20% Intralipid (Sigma-Aldrich, I141) to mimic stromal scattering (μs=2  mm1)15,48 at 600 nm. This wavelength was used as it was the central wavelength in our range of interest (450 to 750 nm). The concentration of Intralipid in water required to simulate the target scattering properties was determined by measuring transmission of phantoms with varying Intralipid concentrations. Then, using the relationship μs=ln(T)L, where L is the path length, a plot of Intralipid scattering properties as a function of Intralipid concentration was obtained (see Appendix). During measurements, the OPRS probe was initially placed in contact with the transparent PDMS layer. Then, the probe was moved away from the PDMS at 50-μm increments until a distance of 2500  μm was reached. OPRS measurements were carried out at each increment. Spectra from each collection fiber were processed as described in Sec. 2.3 and integrated in the wavelength range 450 to 750 nm. The integrated intensities were plotted as a function of Intralipid thickness.

2.5.

Standardization of Clinical Oblique Polarized Reflectance Spectroscopy Measurements

Standardization with respect to the normal tissue was carried out to account for the varying anatomy across patients. The standardization approach was adapted from a method by Rajaram et al.49 wherein it was applied to account for intersubject variations in the skin. A standardization factor was used to account for divergence of scattering from normal sites obtained from a given patient from the overall average intensity of all collected normal spectra. The standardization factor for each patient (Si) was calculated by dividing the integrated intensity of the average spectrum of all collected normal spectra (Nmean) by the integrated intensity of a normal spectrum from each patient (Ni): Si=Nmean/Ni. The Si factor for each patient was used to normalize intensity of all measured normal and abnormal spectra as follows: Mi(λ)=Si×Mi(λ), where Mi(λ) and Mi(λ) represent the standardized and originally measured spectra, respectively. Note that Mi(λ) represents any measured Ii(λ)par and Ii(λ)per spectra. Figure 2 shows OPRS spectra before and after standardization. The standardization scheme retained differences in the spectral shape of the measurements from the normal and abnormal lesions while removing large variations in the overall magnitude caused by interpatient variations and by various anatomical locations in the oral cavity.

Fig. 2

OPRS spectra pre- and poststandardization.

JBO_22_6_065002_f002.png

2.6.

Spectral features

The following eight binary classification tasks were evaluated: (1) normal versus MD, (2) normal versus SD, (3) normal versus MD and SD combined, (4) MD versus SD, (5) benign versus MD, (6) benign versus SD, (7) benign versus MD and SD combined, and (8) benign versus normal. The list of all OPRS features considered in data analysis is shown in Table 1. It includes the spectral mean, which is the average intensity taken across the entire wavelength range (450 to 750 nm), along with the intensity at the most discriminatory wavelengths that were extracted for the following spectra: parallel ()—collected by fibers BF1, BF2, and BF3; perpendicular ()—fibers BF1per, BF2per, and BF3per; diffuse (+)—sums of signals collected by symmetrically positioned fibers collecting parallel and perpendicular spectra; polarization gated ()—differences between parallel signals and corresponding perpendicular signals; parallel/perpendicular (/)—ratios of parallel and perpendicular signals collected by symmetrically positioned fibers; parallel and perpendicular differentials—differences of signals collected by two adjacent fibers; ratios of differential signals; and ratio of polarization gated signals. All these features were extracted from preprocessed and standardized spectra. In addition, the spectra were normalized by the area under the curve (AUC) to emphasize spectral shape differences between various diagnostic categories. For each binary classification task, the most discriminatory wavelength was selected as the wavelength with the highest Welch’s t-statistic value. The Welch’s t-statistic at a given wavelength calculates the absolute difference in mean spectra across patients between the two diagnostic classes relative to the amount of interpatient spectral variation that is observed within the two classes at that wavelength. Another feature that was extracted for all detection fibers was the ratio of the intensities at 576 to 610 nm; this ratio reflects the magnitude of hemoglobin absorption in the scattering spectra.45

Table 1

Individual features that were considered for analysis.

Features
Parallel, I (mean and x nm)BF1
BF2
BF3
Perpendicular, I (mean and x nm)BF1per
BF2per
BF3per
Polarization gated, (mean and x nm)BF1 − BF1per
BF2 − BF2per
BF3 − BF3per
Diffuse, + (mean and x nm)BF1 + BF1per
BF2 + BF2per
BF3 + BF3per
Parallel/perpendicular, / (mean and x nm)BF1/BF1per
BF2/BF2per
BF3/BF3per
Differential, (mean and x nm)BF1 − BF2
BF2 − BF3
BF1 − BF3
Differential, (mean and x nm)BF1per − BF2per
BF2per − BF3per
BF1per − BF3per
Differential ratio, (differential, /differential, ) (mean and x nm)(BF1 − BF2)/(BF1per − BF2per)
(BF2 − BF3)/(BF2per − BF3per)
(BF1 − BF3)/(BF1per − BF3per)
Polarization gated ratio, (mean and x nm)(BF1 − BF1per)/(BF2 − BF2per)
(BF1 − BF1per)/(BF3 − BF3per)
(BF2 − BF2per)/(BF3 − BF3per)
I576  nm/I610  nm, BF1576  nm/610  nm
BF2576  nm/610  nm
BF3576  nm/610  nm
I576  nm/I610  nm, BF1per576  nm/610  nm
BF2per576  nm/610  nm
BF3per576  nm/610  nm

2.7.

Selecting the Most Discriminatory Spectral Features

A total of 120 spectral features were extracted from measured OPRS spectra. These features include 27 spectral means, 27 intensities at the most discriminatory wavelengths, and 6 intensity ratios at 576/610  nm for the total of 60 features for each unnormalized and normalized spectra with the final count of 120 features; this translates to (21201) different possible feature combinations in a two-class classification problem. An exhaustive investigation of all feature combinations would require vast amounts of processing time, and the available data sample size is likely to be insufficient for investigating high-dimensional feature spaces. Therefore, it was necessary to reduce the number of features prior to training classifiers for differentiating between different diagnostic classes. Feature selection was used to identify the most diagnostically relevant features and to reduce redundancies by eliminating features that are closely related to each other. We preferred feature selection over feature extraction (such as principal component analysis) to retain the physical significance of features used for diagnostic classification. Maximum relevance minimum redundancy (mRMR) was employed as the approach for feature selection as mRMR produces a subset of features with the highest relevance (i.e., highest discrimination between the two diagnostic classes) and minimal redundancy (i.e., minimum correlation between features). mRMR has low computational complexity and produces features with smaller classification errors as compared to those obtained from other feature selection strategies.50 Given that sample size is limited for some of the diagnostic classification tasks, leave-one-out cross validation (LOOCV) strategy was used to reduce overtraining. The appropriate set of features from mRMR was calculated as the minimum feature set beyond which the performance on training data does not significantly improve by inclusion of additional features. Upon selecting the most diagnostically relevant features, their performance in discriminating between different diagnostic classes was evaluated by determining their classification accuracy. Random forest classifiers were used in this study owing to their advantages when dealing with small sample sizes and high-dimensional feature space.50 The area under the nonparametric receiver-operating characteristic (ROC) curves (AUC) was calculated to quantify the performance of selected features when combined by the random forest classifier. The random forest classifications along with the determination of the discriminatory wavelengths and feature selection were performed in the R software using the caret and mRMRe packages (R Development Core Team, Vienna, Austria).

2.8.

Testing for Overtraining

To reduce the risk of overtraining associated with the limited number of data in the clinical trial, we employed LOOCV, which leaves out a single subject to generate the random forest models. To further test for overtraining, a permutation test was employed wherein the pathology definition (normal, benign, etc.) was randomly rearranged while preserving the number of cases in each diagnostic category.51 The mean and standard error of the AUCs were obtained from shuffling the pathology definitions 100 times. These values were then compared with the AUC values obtained with correctly assigned diagnostic results to the data.

3.

Results

3.1.

Depth Penetration of Oblique Polarized Reflectance Spectroscopy Collection Fibers

Penetration depths of detection fibers of the OPRS probe were evaluated using an Intralipid phantom simulating stromal scattering. The relationship μs=ln(T)L     between the reduced scattering coefficient, μs (g=0.752)52 at 600 nm, and transmission of Intralipid phantoms at concentrations of 0.05%, 0.1, 0.15%, 0.2%, and 0.25% was used to plot the dependence of the scattering coefficient on the Intralipid concentration (see Appendix). The fit was found to be in good agreement with previously published results.52,53 Using the derived linear fit, the Intralipid concentration of 2.6% in water was found to yield the target μs of 2  mm1. Integrated intensities of scattering signals collected by detection fibers were plotted as a function of phantom thickness as shown in Fig. 3. The signals followed a common trend with an initial increase followed by a saturation. As expected, the intensity profiles of detection fibers positioned further away from the illumination fiber were shifted toward greater depths [Fig. 3(b)]. Also, the perpendicular signals appeared at greater depths as compared to the parallel ones as the result of a gradual depolarization of linear polarized excitation with depth; this trend became less pronounced with increased separation between illumination and detection fibers. The detected signals achieved 90% value of the saturated signal at 750, 900, and 1100  μm for parallel detection fibers BF1, BF2, and BF3, respectively, and at 1200, 1250, and 1400  μm for perpendicular fibers BF1per, BF2per, and BF3per, respectively [Fig. 3(c)]. The polarization gating (BF1 − BF1per, BF1 − BF2per, and BF3 − BF3per) reduced the interrogation depth in the scattering medium as compared to individual collection fibers [Fig. 3(c)].

Fig. 3

(a) A schematic of experimental setup to quantify depth penetration of OPRS collection fibers. (b) Integrated intensities of scattered signal collected by parallel (solid circle) and perpendicular (dotted square) detection fibers as a function of thickness of Intralipid phantom simulating stromal scattering (μs=2  mm1). (c) Depths at 90% signal saturation for individual collection fibers and polarization gated signals.

JBO_22_6_065002_f003.png

3.2.

Sample Distribution of Clinical Measurements

Results of 93 in situ measurements from 25 patients were analyzed. The distribution of anatomical sites measured is shown in Table 2. Representative histological examples of different diagnostic categories, as shown in Fig. 4, illustrate morphological changes associated with the progression of dysplasia in the oral cavity. It is important to note a high degree of similarity between anatomy of benign and MD categories. MD is marked by early dysplastic changes limited to the lower third of the epithelium in the basal and parabasal layers that are not present in the benign category. However, both the benign and the MD images show a well-defined superficial keratinization layer that complicates diagnosis.

Table 2

Distribution of anatomical sites.

NormalBenignMDSD
Buccal8340
Floor2101
Gingiva1210
Mandible0010
Tongue3111189
Total42172410

Fig. 4

Representative images of biopsied tissue sites confirmed as normal, benign, MD, and SD. (scale bar: 200  μm). Tissue slices were stained with H&E for standard histopathological analysis.

JBO_22_6_065002_f004.png

To evaluate the prevalence of keratinization, mean thicknesses of the epithelium and keratin layers for benign, MD, and SD diagnostic categories were measured in scanned H&E stained slides of biopsied samples using Panoramic Viewer (3DHISTECH) (Fig. 5). Mean epithelial thicknesses (epithelium + keratin) for benign, MD, and SD were ca. 426, 504, and 701  μm, respectively, indicating thickening of the epithelial layer with dysplasia progression. Lesions in the oral cavity often experience hyperkeratosis due to chronic irritation, which is reflected in an increased degree of keratinization.54

Fig. 5

Mean epithelial and keratin thicknesses of biopsied sites (benign, MD, and SD).

JBO_22_6_065002_f005.png

3.3.

Polarized Reflectance Spectra

Averaged parallel (), perpendicular (), diffuse (+), and polarization gated () reflectance spectra for each diagnostic category are shown in Fig. 6(b). Qualitative analyses of the spectra uncovered substantial differences between diagnostic categories. These differences were wavelength dependent and were quite substantial in certain narrow wavelength regions that prompted analysis of the most discriminatory wavelengths and wavelength regions that are described below. Overall, average benign spectra tended to have the highest intensity as compared to other diagnostic categories likely due to a high degree of keratinization associated with benign oral cavity lesions. The mean abnormal spectra (MD and SD) exhibited larger intensities than the normal spectra that can be attributed to higher scattering due to keratinization of these sites and to morphological changes associated with dysplasia, such as increased nuclear size, hyperchromasia, and pleomorphism.55 The spectra were also normalized by the AUC to emphasize shape differences between diagnostic categories (Fig. 7). It is interesting to note that the normalization revealed small but distinct spectral differences between all diagnostic categories in the parallel spectra; however, in the perpendicular spectra, MD and SD categories are virtually indistinguishable, whereas normal and benign spectra are clearly separated.

Fig. 6

Averaged spectra for each diagnostic category for parallel (), perpendicular (), diffuse (+), and polarization gated () reflectance spectra.

JBO_22_6_065002_f006.png

Fig. 7

Polarized reflectance spectra normalized to the AUC; all collected spectra for each diagnostic category were first normalized by the AUC and then averaged.

JBO_22_6_065002_f007.png

3.4.

Diagnostically Significant Spectral Features

Figure 8 shows a diagram that summarizes the most discriminatory wavelengths that provide the maximum separation between each binary diagnostic classification class. Each tickmark on the x-axis represents a spectral feature of interest listed in Table 1—a total of 27 features for each binary classification; the corresponding discriminatory wavelengths are identified by dots along the y-axis. The analysis was carried out for unnormalized [Fig. 8(a)] and AUC normalized spectra [Fig. 8(b)]. The full list of discriminatory wavelengths can be found in Appendix, Tables 7 and 8.

Fig. 8

Diagrams illustrating the most discriminatory wavelengths for all features of interest for each feature binary classification task for unnormalized (a) and AUC normalized (b) spectra. Spectral features of interest are indicated by tickmarks along x-axis—a total of 27 marks for each classification task; the y-axis shows wavelengths.

JBO_22_6_065002_f008.png

To determine spectral regions with the most diagnostic relevance, the entire wavelength range (450 to 750 nm) was divided into 20 nm spectral bands and the frequency of appearance of discriminatory wavelengths in each band was evaluated (Table 3). For both unnormalized and area normalized spectra, wavelengths associated with hemoglobin absorbance between 510 and 610 nm appeared frequently in discrimination between normal and benign versus dysplasia (MD and SD) indicating that blood absorption plays an important role in discrimination of histologically normal and abnormal tissue. The wavelength band of 450 to 469 nm was also prominent in all diagnostic categories for unnormalized features and especially in the separation of benign from MD and SD. This trend can be attributed to a strong superficial scattering of light in the blue spectral range from the surface keratin layer. Indeed, average spectra for the benign category shown in Fig. 6 generally exhibited a steeper, more negative slope in the wavelength range 450 to 500 nm as compared to other diagnostic categories. However, the wavelength band of 450 to 469 nm did not appear frequently for the area-normalized spectra indicating that this wavelength region was strongly associated with signal amplitude rather than spectral shape. Prevalent discriminatory wavelengths for the discrimination of MD from SD were not associated with hemoglobin absorption indicating that this classification was more sensitive to differences in scattering. Indeed, the wavelength region above 600 nm is highly significant in binary tasks associated with keratinized tissues, such as benign, MD, and SD. It could be associated with better penetration of red-NIR light through the keratinized layer in these sites that results in the collection of more diagnostically relevant information.

Table 3

Wavelengths regions with the greatest appearance for each classification task that are listed in order of appearance frequency. Only frequencies above 10% are listed.

Unnormalized spectraNormalized spectra
Wavelengths (nm) Appearance (%)Wavelengths (nm) Appearance (%)
Normal fromSD450 to 46937510 to 52930
510 to 52915450 to 46922
530 to 54915590 to 60922
730 to 75011
MD530 to 54937530 to 54922
590 to 60919550 to 56922
490 to 50919570 to 58911
450 to 46911590 to 60911
690 to 70911
MD and SD450 to 46922550 to 56922
490 to 50922570 to 58922
530 to 54922590 to 60915
550 to 56911510 to 52911
590 to 60911
Benign fromSD450 to 46967510 to 52926
570 to 58919530 to 54919
590 to 60915
610 to 62915
570 to 58911
MD450 to 46944610 to 62930
570 to 58930470 to 48919
590 to 60915
630 to 64911
MD and SD450 to 46948610 to 62956
570 to 58922570 to 58915
630 to 64911630 to 64911
Normal450 to 46959610 to 62944
570 to 58915630 to 64915
MD fromSD450 to 46933610 to 62937
730 to 75026630 to 64926
470 to 48922730 to 75011

After determining the most discriminatory wavelengths, we carried out selection with mRMR that included all features listed in Table 1 for all detection fibers to identify the best combination of spectral features for each diagnostic classification task (Table 4). Features selected by the mRMR algorithm included scattering signals from all detection fibers as well as features associated with polarization gated and diffuse scattering spectra; this implies that scattering signals collected from various depths were found to be diagnostically relevant. The discrimination of normal tissue from dysplasia (SD, MD, and MD and SD combined) and the benign category produced excellent results with AUC values close to 1 (Table 4 and Fig. 9). However, the discrimination of the more clinically challenging classification of the benign from dysplasia was more challenging as reflected by significantly lower AUCs. The poorer performance in discriminating the benign and dysplastic categories could be attributed to keratinization confounding the diagnosis. Indeed, an increase in scattering is expected for MD and SD cases due to alterations in epithelial morphology, such as increased nuclear size, hyperchromasia, and pleomorphism along with keratin formation. However, benign sites are also associated with an increased scattering signal due to a high degree of keratinization (Fig. 4). A strong scattering by the keratin layer can create a significant background scattering, thus, interfering with detection of scattering signatures of dysplasia.

Table 4

Features found to be important in discriminating between diagnostic categories using mRMR and the corresponding AUC. Features identified with a wavelength (nm) refer to intensity at the most discriminatory wavelength, and those labelled with norm are associated with normalized spectra. No wavelengths are listed for spectral features associated with the spectral mean—the average intensity taken across the entire wavelength range.

FeaturesAUC
Normal fromSDBF3, norm., 513 nm0.998
BF2per − BF3per, 545 nm
BF3 + BF3per, 513 nm
BF1, norm., 510 nm
MDBF2per, norm., 551 nm0.988
BF3/BF3per, 540 nm
(BF2 − BF2per)/(BF3 − BF3per), 544 nm
(BF1 − BF2)/(BF1per − BF2per)
(BF1 − BF2)/(BF1per − BF2per), 543 nm
MD and SDBF2per, norm., 552 nm0.999
BF2 − BF3, 546 nm
BF2per − BF3per, 554 nm
BF3, norm., 510 nm
Benign fromSD(BF2 − BF3)/(BF2per − BF3per), 453 nm0.850
MDBF3, 453 nm0.822
MD and SDBF3 + BF3per, 450 nm0.744
BF2/BF2per
BF1, norm., 587 nm
(BF1 − BF1per)/(BF2 − BF2per)
BF2per − BF3per, 450 nm
(BF2 − BF2per)/(BF3 − BF3per), 639 nm
NormalBF3, norm., 632 nm0.980
BF1per − BF2per, 577 nm
(BF2 − BF2per)/(BF3 − BF3per), 639 nm
MD fromSDBF3/BF3per, 450 nm0.748
BF3per, norm., 602 nm
(BF1 − BF2)/(BF1per − BF2per)
BF2per − BF3per, 750 nm

Fig. 9

ROC curves for each classification task.

JBO_22_6_065002_f009.png

After conducting the global analysis with all detection fibers, we addressed the question of whether the multifiber design of this OPRS probe improves diagnostic performance as compared to any given single fiber pair of the probe. Similar to the multifiber analysis, the most diagnostically relevant spectral features and the corresponding AUCs for each parallel/perpendicular pair of detection fibers were determined using feature selection with mRMR (Table 5). With global analysis, including all detection fibers, the separation of normal from MD and SD resulted in excellent AUCs for all fiber pairs with no significant difference between different pairs or their combination. However, discriminations of the benign from MD and from SD categories significantly worsened as compared to the combination of all fibers. These results were further supported by the statistical overtraining analysis described below.

Table 5

Features associated with individual fiber pairs found to be important in discriminating diagnostic groups using mRMR and the corresponding AUCs.

BF1 and BF1perAUCBF2 and BF2perAUCBF3 and BF3perAUC
Normal fromSDBF1, norm., 510 nm0.993BF2, norm., 514 nm1.000BF3, norm., 513 nm1.000
BF1 + BF1per, 544 nmBF3/BF3per, 450 nm
BF1/BF1per, 450 nmBF3 + BF3per, 513 nm
BF1BF3per, norm., 595 nm
BF1per, norm., 562 nm
MDBF1per, norm., 553 nm0.956BF2, norm., 703 nm0.977BF3per, norm., 552 nm0.989
BF1/BF1per, 542 nmBF3/BF3per, 507 nm
BF1BF3per576  nm/610  nm
MD and SDBF1per, norm., 554 nm0.999BF2per, norm., 552 nm0.998BF3per, norm., 554 nm1.000
BF1BF2BF3/BF3per, 450 nm
BF1, norm., 511 nmBF2/BF2perBF3, norm., 510 nm
BF2, norm., 513 nm
Benign fromSDBF1 − BF1per, 581 nm0.606BF20.588BF3/BF3per, 450 nm0.713
BF1per, 450 nm
BF1per, norm., 612 nm
BF1
MDBF1, 582 nm0.733BF2, 450 nm0.745BF3, 450 nm0.688
BF1/BF1perBF2/BF2perBF3 − BF3per
BF1, norm., 616 nm
BF1 − BF1per, 582 nm
MD and SDBF1 + BF1per, 450 nm0.669BF2, 450 nm0.731BF3 + BF3per, 450 nm0.612
BF1/BF1perBF2/BF2per
BF1, norm., 587 nm
BF1per, norm., 614 nm
BF1
BF1per, 450 nm
NormalBF1 + BF1per, 450 nm0.983BF2, 450 nm0.982BF3, norm., 632 nm0.965
BF1/BF1perBF2per, norm., 628 nm
BF1per, norm., 450 nm
MD fromSDBF1 − BF1per, 450 nm0.658BF2 − BF2per, 450 nm0.635BF3/BF3per, 450 nm0.715
BF1per, norm., 611 nmBF2per, norm., 610 nmBF3per, norm., 602 nm
BF1576  nm/610  nmBF3, 485 nm
BF1, norm., 516 nm

3.5.

Check for Overtraining

LOOCV together with a permutation test wherein the data were randomly assigned a diagnostic class was used to check for overtraining. The results from the permutation test are shown in Fig. 10. The mean±standard deviation of the AUCs for the data with randomly shuffled diagnoses are shown as black lines and are compared with real AUCs from Tables 4 and 5 obtained for the correctly classified dataset. The real AUCs in classifications of the normal from the benign and dysplastic cases were well above the errors bars of random permutation tests confirming statistical significance of data analyses. However, there was no significant difference between data analyses that included all fibers or any separate individual parallel/perpendicular detection fiber pair. The benign versus MD and SD classification tasks were also statistically significant in the case of the global analysis that included all detection fibers, but it was not the case for any of the individual detection fiber pairs. The real AUCs of the MD versus SD classification task overlap with the error bars of the shuffled AUCs that indicates overtraining of the dataset. Therefore, the classification task failed to discriminate MD and SD sites.

Fig. 10

Permutation test to check for overtraining. Collected spectra were randomly assigned diagnoses while keeping the overall distribution of diagnostic categories the same; samples were randomly shuffled 100 times. The mean and standard deviation of the AUC obtained from the shuffling are presented and compared to the real AUCs shown as stars (all fibers combined), dots (BF1 fiber pair), squares (BF2), and circles (BF3).

JBO_22_6_065002_f010.png

We also performed analyses using a dataset that did not include features extracted from area-normalized spectra (Appendix, Fig. 12). It is interesting to note that no statistical significance was achieved in discrimination of the benign from MD and SD categories using only unnormalized data.

4.

Discussion and Conclusions

The challenge of distinguishing malignant lesions from benign, inflammatory conditions in the oral cavity using optical methods is well known. For example, there are two commercially available systems based on tissue reflectance—Vizilite Plus (DenMat, Lompoc, California) and Microlux DL (AdDent, Danbury, Connecticut)—that utilize an acetic acid wash to enhance scattering from abnormal lesions due to the effect of acetowhitening.56,57 The sensitivity of Vizilite Plus was reported to be 100%; however, the specificity is poor ranging from 0% to 14%.58 Microlux DL fared better with a sensitivity of 78% and a specificity of 71%, but its positive predictive value was just 0.37 in a clinical trial of patients with oral white lesions.59

A number of research groups evaluated spectroscopic techniques in diagnosis of malignancies in the oral cavity with various degree of success. Amelink et al.60 developed optical fiber probes for a quantitative assessment of oral tissue with differential path-length spectroscopy (DPS); they found that DPS can discriminate nondysplastic and dysplastic leukoplakias, which are characterized by a high level of keratinization, with 91% sensitivity and 80% specificity. Müller et al.22 reported a sensitivity of 64% and a specificity of 90% in discriminating dysplastic from cancerous cases in a clinical study that combined fluorescence, diffuse reflectance, and light scattering spectroscopy. In a study of a ball lens coupled optical probe that collected autofluorescence and diffuse reflectance of oral tissue from different depths, Schwarz et al.10 observed a sensitivity of 82% and specificity of 87% when discriminating normal combined with MD versus moderate dysplasia plus cancer cases in 119 nonkeratinized sites; however, the sensitivity and specificity were down to 79% and 80%, respectively, when a dataset of 114 keratinized sites was analyzed.

There have been a number of very interesting clinical studies in the literature reporting the use of physical models in the detection of neoplastic changes in human patients. For example, Perelman’s group used an algorithm for extraction of nuclear sizes from scattering spectra collected using endoscopic polarized scanning spectroscopy (EPSS) in patients with Barrett’s esophagus to identify early malignant changes.61 Backman et al.62 developed an elegant method called low-coherence enhanced backscattering spectroscopy to quantify morphological changes in epithelial tissue associated with carcinogenesis; this technology showed sensitivity of 88% and 71% in advanced adenomas and nonadvanced adenomas, respectively.63 Furthermore, it was demonstrated that quantitation of changes in blood supply using physical models of polarization gated spectroscopy can differentiate pancreatic adenocarcinoma from normal tissue with 92% sensitivity and 86% specificity.2 However, despite been very promising these technologies are better suited for unkeratinized thin epithelial tissue. Alternatively, physical models developed for diffuse reflectance spectroscopy are designed for an assessment of deeper located tumors such as breast lesions.64 These models were used to evaluate changes in tissue concentration of oxy- and deoxyhemoglobin, water, and lipid;64 this technology showed promise in monitoring of response to breast cancer neoadjuvant therapy.65 However, physical models, which can accurately consider the influence of a strong scattering from keratinized tissues, are still lacking.

In our group’s previous clinical trial using the OPRS fiber probe with a single pair of parallel/perpendicular detection fibers, we obtained statistically significant discrimination between the benign category and SD lesions with 100% sensitivity and 85% specificity and between the benign and MD sites with 92% sensitivity and 69% specificity.24 We also observed a sensitivity of 90% and specificity of 86% in the separation of normal from SD and a sensitivity of 75% and specificity of 73% in the separation of normal from MD.24 The multifiber OPRS probe presented here was inspired by our phantom experiments in well-defined multilayer scattering phantoms mimicking dysplasia of the epithelial tissue, which showed that a multifiber probe design can improve depth resolved spectroscopic measurements in a scattering media;45 however, these phantom studies did not take into account a strong surface scattering by a keratin layer.

The sensitivity and specificity values of OPRS for all diagnostic classification tasks are listed in Table 6. ROC thresholds were chosen such that both high sensitivity and specificity were produced. The discrimination of normal from benign, MD, SD, and MD/SD combined generated sensitivity and specificity values above 95%. These results are a big improvement over our first clinical trial. It is interesting to note that there was no significant difference in the performance of all fibers and any particular fiber pair in these classification tasks. As expected, discrimination of benign from dysplasia resulted in lower specificities and sensitivities values. However, here the analyses including multiple detection fibers performed significantly better as compared to individual fiber pairs in discriminating benign from MD and benign from SD where sensitivity and specificity of 71%/93% and 80%/93%, respectively, were achieved. Furthermore, in these two classification tasks, the overtraining test was passed only when an analysis was performed using data from all detection fibers as shown in Fig. 9. Therefore, the multifiber OPRS probe design improved discrimination of benign from dysplastic lesions in the oral cavity as compared to a single fiber pair. It is also important to note that normalizing reflectance spectra to the AUC was critical in achieving statistically significant classification of benign versus MD and SD categories as can be seen from comparing permutation tests for normalized (Fig. 10) and unnormalized (Appendix, Fig. 12) data.

Table 6

Sensitivity and specificity of OPRS for all diagnostic classification tasks. The values that pass the statistical overtraining test are highlighted in bold font.

All fibersBF1 and BF1perBF2 and BF2perBF3 and BF3per
SensitivitySpecificitySensitivitySpecificitySensitivitySpecificitySensitivitySpecificity
Normal fromSD100981009390100100100
MD96951009395989298
MD and SD10010010010097100100100
Benign fromSD8093805390537080
MD7193757675765487
MD and SD7480478285716860
Normal100939594939410093
MD fromSD8071507960716083

The decrease in the performance in discrimination between benign, MD, and SD diagnostic categories is most likely associated with an overwhelming scattering from the epithelial keratin layer that is present in all of these lesions (Fig. 5). Furthermore, variations in thickness of the keratin layer result in a high variability of scattering spectra in benign category even after the standardization procedure as can be seen in Fig. 2. These thickness variations are unpredictable and do not have a diagnostic value that significantly complicates optical detection of dysplasia in keratinized lesions. The same observation was previously made in studies of dysplastic and nondysplastic oral leukoplakias by Amelink et al.60

Here, we used intrinsic data collected by the multifiber OPRS probe to show that multiple detection fibers improved results of binary classification tasks, which include benign lesions. It is interesting to note that there were noticeable differences in the performance of the multifiber OPRS probe and the one by the single detection fiber pair OPRS probe evaluated in our previous pilot clinical trial;24 these include significantly better discrimination between normal and all other diagnostic categories in the current study and a better performance of the previously reported probe in discriminating benign and SD. The observed differences in the performance are likely associated with changes in probe design as well as with variations between patient groups that were evaluated in these two studies.

Moreover, these results likely reflect challenges in achieving an adequate depth resolution of the spectroscopic measurements. Indeed, the axial extent of overlap volumes between the probe’s illumination and collection cones ranged from 750 to 1100  μm and from 1200 to 1400  μm for parallel and perpendicular detection fibers, respectively, [Fig. 3(c)] that is significantly larger than the measured epithelial thickness, which was varying from ca. 426 to 701  μm in progression from benign to SD (Fig. 5). Consequently, there was a strong contribution of the stromal scattering in collected spectra that was confounded by a strong keratin scattering in the case of benign and dysplastic sites.

It is important to note that our standardization and normalization scheme does not account for the diverse anatomy of the oral cavity. McGee et al.66 evaluated the impact of anatomy on classification accuracy in a study of 710 spectra from nine different anatomic sites in the oral cavity and found that significant dissimilarities in spectral features exist between different anatomic sites. In a later study wherein McGee et al. used anatomic site specific algorithms, a sensitivity and specificity of only 53% and 70%, respectively, were achieved in the discrimination of benign from dysplasia when data were grouped together regardless of anatomic site.44 However, the sensitivity and specificity improved to 92% and 67%, respectively, when only sites from the lateral surface of the tongue were considered. A similar trend was observed with a sensitivity of 94% and specificity of 60% when samples from the floor of the mouth and ventral surface of the tongue, anatomic sites found to have similar spectral properties, were analyzed. In future OPRS studies, samples could be stratified by an anatomic site, which could improve diagnostic accuracy. However, due to the limited sample size in this study, stratification was not possible.

There are also some limitations that need to be taken into account with respect to our study. Although every effort was made to carry out consistent spectral measurements, the pressure applied to the probe was not controlled in a quantitative way. Although some studies indicated that probe pressure does not significantly alter spectroscopic measurements,67 other reports showed that a firm pressure can result in measurable spectral changes.68 Therefore, the lack of precise control could have added an experimental variable to our data. In addition, it is conceivable that more than one wavelength per spectral feature could be meaningful in the discrimination of different classes. However, an in-depth investigation would entail considering all 301 wavelengths (from 450 to 750 nm) as possible features and conducting feature selection to narrow down to a small set of wavelengths that best discriminates any two diagnostic classes. While this would be a very interesting question to investigate, the sample size in this study was not enough for such an investigation because of a high probability of over-training.

In summary, the multifiber OPRS probe design allows simultaneous implementation of a number of approaches to enhancement of depth sensitivity of scattering measurements in a turbid media including polarization gating, varying source–detector separation and differential spectroscopy. Our study has shown that this combination can discriminate benign and dysplastic lesions in the context of highly keratinized epithelium of the oral cavity. However, a strong scattering from the keratin layer that is characterized by a high degree of interpatient variations remains a significant challenge. A number of modifications can be envisioned for multifiber OPRS in order to further improve its performance in keratinized tissues including the optimization of probing depth of detection fibers and implementation of physical models of light interactions with various tissue morphologies.34,37,6970.71 However, the results of this clinical study and our previous phantom experiments45 indicate that multifiber OPRS could be better suited for epithelial tissues with a low amount of keratinization such as cervix, colon, or bladder. There are also interesting developments in depth-resolved spectroscopic measurements in tissue including spectroscopic OCT72,73 and depth-resolved reflectance spectroscopy using elliptically polarized light.7475.76 These emerging methods could offer new solutions to optical detection of dysplasia in keratinized tissues.

Appendices

Appendix

The section includes the graph from which the concentration of Intralipid required to produce the target scattering coefficient of 2  mm1 at 600 nm was obtained (Fig. 11); Tables 7 and 8 with a list of the best discriminatory wavelengths for all unnormalized and normalized, respectively, spectral features analyzed in this study; and Fig. 12 with the results of the permutation test for unnormalized spectral features.

Fig. 11

Reduced scattering coefficient values (μs, mm1) as a function of Intralipid concentrations.

JBO_22_6_065002_f011.png

Table 7

Wavelengths that provided the maximum separation between different binary diagnostic classes for unnormalized spectra.

Normal fromBenign fromMD from
SDMDMD and SDSDMDMD and SDNormalSD
BF1542541538582582582450481
BF2564557561450450450450485
BF3514507509450450450450485
BF1per750604506450450450450750
BF2per564605599450450450450750
BF3per595604599450650450450750
BF1 − BF1per450450450581582582582450
BF2 − BF2per450708572581583583582450
BF3 − BF3per515508509450450450450483
BF1 + BF1per544542537450450450450521
BF2 + BF2per512504509450450450450750
BF3 + BF3per513507509450450450450515
BF1/BF1per450542450450554734737450
BF2/BF2per450542543450549728727450
BF3/BF3per450540450450589701596450
BF1 − BF2450450450582582582450480
BF2 − BF3543547546450450450450750
BF1 − BF3450450450581581581450480
BF1per − BF2per657557542674579584577656
BF2per − BF3per545603597450450450450750
BF1per − BF3per750602554450450450450750
(BF1 − BF2)/(BF1per − BF2per)458543497659579624583455
(BF2 − BF3)/(BF2per − BF3per)750546467665609640641450
(BF1 − BF3)/(BF1per − BF3per)450544733450604738690450
(BF1 − BF1per)/(BF2 − BF2per)604572572463453453459583
(BF1 − BF1per)/BF3 − BF3per)582508553646633646661702
(BF2 − BF2per)/(BF3 − BF3per)453544541453639639639453

Table 8

Wavelengths that provided the maximum separation between different binary diagnostic classes for normalized spectra.

Normal fromBenign fromMD from
SDMDMD and SDSDMDMD and SDNormalSD
BF1562553554611612620614626
BF2511551552610612618619628
BF3595552554602617617617627
BF1per510702511516589616587623
BF2per514703513517613621621628
BF3per513701510515615616616632
BF1 − BF1per458544579697487584487482
BF2 − BF2per450543543585485590450497
BF3 − BF3per463540450581480588452594
BF1 + BF1per515555556516612620620633
BF2 + BF2per512553560514612620620628
BF3 + BF3per511553556514608617617629
BF1/BF1per450544579547489585732737
BF2/BF2per450544576547490590728738
BF3/BF3per463540580547487589731743
BF1 − BF2595590592604632629630636
BF2 − BF3750671690750630636637640
BF1 − BF3595594593750625631630638
BF1per − BF2per595582587602603623624628
BF2per − BF3per601580604543606613607623
BF1per − BF3per596585592600601622609623
(BF1 − BF2)/(BF1per − BF2per)733686693622568724572450
(BF2 − BF3)/(BF2per − BF3per)652527450540634537617656
(BF1 − BF3)/(BF1per − BF3per)573498534580583491491665
(BF1 − BF1per)/(BF2 − BF2per)524659661529653647648523
(BF1 − BF1per)/(BF3 − BF3per)643450500629450621613635
(BF2 − BF2per)/(BF3 − BF3per)641600579639544627625636

Fig. 12

Permutation test to check for overtraining with unnormalized spectral features. Samples were randomly shuffled 100 times. The mean and standard deviation of the AUC obtained from the shuffling are presented and compared to the real AUCs. Note that analyses with this unnormalized spectral dataset failed to achieve statistical significance in binary classification tasks of benign versus MD and benign versus SD.

JBO_22_6_065002_f012.png

Disclosures

Authors have no financial interests associated with the results of this study.

Acknowledgments

This work was supported in part by the National Institutes of Health (NIH) under Grant Nos. R01 EB003540 and R01 CA103830. Maria Bailey was also supported in part by the NIH under a T32 training grant (No. EB007507). We also would like to thank Mr. Bobby Knight for his contributions to manufacturing of the initial prototypes of the OPRS probes. Clinical trials have been approved by the Institutional Review Boards at the UT M.D. Anderson Cancer Center and British Columbia Cancer Agency.

References

1. N. A. Howlader et al., Eds., “SEER cancer statistics review, 1975–2010,” National Cancer Institute, Bethesda, Maryland, 2012,  http://seer.cancer.gov/csr/1975_2010/ (April 2013). Google Scholar

2. M. Patel et al., “Polarization gating spectroscopy of normal-appearing duodenal mucosa to detect pancreatic cancer,” Gastrointest. Endoscopy 80(5), 786–793 (2014).GAENBQ0016-5107 http://dx.doi.org/10.1016/j.gie.2014.03.031 Google Scholar

3. P. Brocklehurst et al., “Screening programmes for the early detection and prevention of oral cancer,” Cochrane Database Syst. Rev. 11, CD004150 (2013). http://dx.doi.org/10.1002/14651858.CD004150.pub4 Google Scholar

4. J. Myers, Oral Cancer Metastasis, Springer Science & Business Media, New York (2009). Google Scholar

5. B. Swinson et al., “Optical techniques in diagnosis of head and neck malignancy,” Oral Oncol. 42(3), 221–228 (2006).EJCCER1368-8375 http://dx.doi.org/10.1016/j.oraloncology.2005.05.001 Google Scholar

6. P. Wilder-Smith et al., “Optical diagnostics in the oral cavity: an overview,” Oral Dis. 16(8), 717–728 (2010). http://dx.doi.org/10.1111/odi.2010.16.issue-8 Google Scholar

7. L. Fass, “Imaging and cancer: a review,” Mol. Oncol. 2(2), 115–152 (2008). http://dx.doi.org/10.1016/j.molonc.2008.04.001 Google Scholar

8. C. G. Hadjipanayis et al., “Current and future clinical applications for optical imaging of cancer: from intraoperative surgical guidance to cancer screening,” Semin. Oncol. 38, 109–118 (2011).SOLGAV0093-7754 http://dx.doi.org/10.1053/j.seminoncol.2010.11.008 Google Scholar

9. T. Upile et al., “Head & neck optical diagnostics: vision of the future of surgery,” Head Neck Oncol. 1(1), 1 (2009). http://dx.doi.org/10.1186/1758-3284-1-1 Google Scholar

10. R. A. Schwarz et al., “Noninvasive evaluation of oral lesions using depth-sensitive optical spectroscopy,” Cancer 115(8), 1669–1679 (2009).CANCAR0008-543X http://dx.doi.org/10.1002/cncr.v115:8 Google Scholar

11. P. M. Lane et al., “Simple device for the direct visualization of oral-cavity tissue fluorescence,” J. Biomed. Opt. 11(2), 024006 (2006).JBOPFO1083-3668 http://dx.doi.org/10.1117/1.2193157 Google Scholar

12. K. Sokolov, M. Follen and R. Richards-Kortum, “Optical spectroscopy for detection of neoplasia,” Curr. Opin. Chem. Biol. 6(5), 651–658 (2002).COCBF41367-5931 http://dx.doi.org/10.1016/S1367-5931(02)00381-2 Google Scholar

13. K. Sokolov et al., “Realistic three-dimensional epithelial tissue phantoms for biomedical optics,” J. Biomed. Opt. 7(1), 148–156 (2002).JBOPFO1083-3668 http://dx.doi.org/10.1117/1.1427052 Google Scholar

14. I. Pavlova et al., “Microanatomical and biochemical origins of normal and precancerous cervical autofluorescence using laser-scanning fluorescence confocal microscopy,” Photochem. Photobiol. 77(5), 550–555 (2003).PHCBAP0031-8655 http://dx.doi.org/10.1562/0031-8655(2003)077<0550:MABOON>2.0.CO;2 Google Scholar

15. R. Drezek et al., “Understanding the contributions of NADH and collagen to cervical tissue fluorescence spectra: modeling, measurements, and implications,” J. Biomed. Opt. 6(4), 385–396 (2001).JBOPFO1083-3668 http://dx.doi.org/10.1117/1.1413209 Google Scholar

16. M. A. Huber, “Assessment of the VELscope as an adjunctive examination tool,” Tex. Dent. J. 126(6), 528–535 (2009). Google Scholar

17. K. Awan, P. Morgan and S. Warnakulasuriya, “Evaluation of an autofluorescence based imaging system (VELscope™) in the detection of oral potentially malignant disorders and benign keratoses,” Oral Oncol. 47(4), 274–277 (2011).EJCCER1368-8375 http://dx.doi.org/10.1016/j.oraloncology.2011.02.001 Google Scholar

18. M. Olivo, R. Bhuvaneswari and I. Keogh, “Advances in bio-optical imaging for the diagnosis of early oral cancer,” Pharmaceutics 3(3), 354–378 (2011). http://dx.doi.org/10.3390/pharmaceutics3030354 Google Scholar

19. R. Malini et al., “Discrimination of normal, inflammatory, premalignant, and malignant oral tissue: a Raman spectroscopy study,” Biopolymers 81(3), 179–193 (2006).BIPMAA0006-3525 http://dx.doi.org/10.1002/(ISSN)1097-0282 Google Scholar

20. C. M. Krishna et al., “Micro-Raman spectroscopy for optical pathology of oral squamous cell carcinoma,” Appl. Spectrosc. 58(9), 1128–1135 (2004).APSPA40003-7028 http://dx.doi.org/10.1366/0003702041959460 Google Scholar

21. W. Jerjes et al., “Detection of cervical intranodal metastasis in oral cancer using elastic scattering spectroscopy,” Oral Oncol. 40(7), 673–678 (2004).EJCCER1368-8375 http://dx.doi.org/10.1016/j.oraloncology.2004.01.009 Google Scholar

22. M. G. Müller et al., “Spectroscopic detection and evaluation of morphologic and biochemical changes in early human oral carcinoma,” Cancer 97(7), 1681–1692 (2003).CANCAR0008-543X http://dx.doi.org/10.1002/cncr.v97:7 Google Scholar

23. A. Sharwani et al., “Assessment of oral premalignancy using elastic scattering spectroscopy,” Oral Oncol. 42(4), 343–349 (2006).EJCCER1368-8375 http://dx.doi.org/10.1016/j.oraloncology.2005.08.008 Google Scholar

24. L. T. Nieman et al., “Probing local tissue changes in the oral cavity for early detection of cancer using oblique polarized reflectance spectroscopy: a pilot clinical trial,” J. Biomed. Opt. 13(2), 024011 (2008).JBOPFO1083-3668 http://dx.doi.org/10.1117/1.2907450 Google Scholar

25. R. A. Schwarz et al., “Ball lens coupled fiber-optic probe for depth-resolved spectroscopy of epithelial tissue,” Opt. Lett. 30(10), 1159–1161 (2005).OPLEDP0146-9592 http://dx.doi.org/10.1364/OL.30.001159 Google Scholar

26. Q. Liu and N. Ramanujam, “Relationship between depth of a target in a turbid medium and fluorescence measured by a variable-aperture method,” Opt. Lett. 27(2), 104–106 (2002).OPLEDP0146-9592 http://dx.doi.org/10.1364/OL.27.000104 Google Scholar

27. T. J. Pfefer et al., “Selective detection of fluorophore layers in turbid media: the role of fiber-optic probe design,” Opt. Lett. 28(2), 120–122 (2003).OPLEDP0146-9592 http://dx.doi.org/10.1364/OL.28.000120 Google Scholar

28. T. J. Pfefer et al., “Multiple-fiber probe design for fluorescence spectroscopy in tissue,” Appl. Opt. 41(22), 4712–4721 (2002).APOPAI0003-6935 http://dx.doi.org/10.1364/AO.41.004712 Google Scholar

29. Z. Changfang, L. Quan and N. Ramanujam, “Effect of fiber optic probe geometry on depth-resolved fluorescence measurements from epithelial tissues: a Monte Carlo simulation,” J. Biomed. Opt. 8(2), 237–247 (2003).JBOPFO1083-3668 http://dx.doi.org/10.1117/1.1559058 Google Scholar

30. A. Amelink et al., “In vivo measurement of the local optical properties of tissue by use of differential path-length spectroscopy,” Opt. Lett. 29(10), 1087–1089 (2004).OPLEDP0146-9592 http://dx.doi.org/10.1364/OL.29.001087 Google Scholar

31. O. R. Scepanovic et al., “A multimodal spectroscopy system for real-time disease diagnosis,” Rev. Sci. Instrum. 80(4), 043103 (2009).RSINAK0034-6748 http://dx.doi.org/10.1063/1.3117832 Google Scholar

32. L. Nieman et al., “Optical sectioning using a fiber probe with an angled illumination-collection geometry: evaluation in engineered tissue phantoms,” Appl. Opt. 43(6), 1308–1319 (2004).APOPAI0003-6935 http://dx.doi.org/10.1364/AO.43.001308 Google Scholar

33. A. M. J. Wang et al., “Depth-sensitive reflectance measurements using obliquely oriented fiber probes,” J. Biomed. Opt. 10(4), 044017 (2005).JBOPFO1083-3668 http://dx.doi.org/10.1117/1.1989335 Google Scholar

34. D. Arifler et al., “Reflectance spectroscopy for diagnosis of epithelial precancer: model-based analysis of fiber-optic probe designs to resolve spectral information from epithelium and stroma,” Appl. Opt. 44(20), 4291–4305 (2005).APOPAI0003-6935 http://dx.doi.org/10.1364/AO.44.004291 Google Scholar

35. R. Reif, O. A’Amar and I. J. Bigio, “Analytical model of light reflectance for extraction of the optical properties in small volumes of turbid media,” Appl. Opt. 46(29), 7317–7328 (2007).APOPAI0003-6935 http://dx.doi.org/10.1364/AO.46.007317 Google Scholar

36. A. Garcia-Uribe et al., “Skin cancer detection by spectroscopic oblique-incidence reflectometry: classification and physiological origins,” Appl. Opt. 43(13), 2643–2650 (2004).APOPAI0003-6935 http://dx.doi.org/10.1364/AO.43.002643 Google Scholar

37. K. Sokolov et al., “Reflectance spectroscopy with polarized light: is it sensitive to cellular and nuclear morphology,” Opt. Express 5(13), 302–317 (1999).OPEXFF1094-4087 http://dx.doi.org/10.1364/OE.5.000302 Google Scholar

38. V. Backman et al., “Polarized light scattering spectroscopy for quantitative measurement of epithelial cellular structures in situ,” IEEE J. Sel. Top. Quantum Electron. 5(4), 1019–1026 (1999).IJSQEN1077-260X http://dx.doi.org/10.1109/2944.796325 Google Scholar

39. R. R. Anderson, “Polarized light examination and photography of the skin,” Arch. Dermatol. 127(7), 1000–1005 (1991).ARDEAC0003-987X http://dx.doi.org/10.1001/archderm.1991.01680060074007 Google Scholar

40. S. L. Jacques, J. R. Roman and K. Lee, “Imaging superficial tissues with polarized light,” Lasers Surg. Med. 26, 119–129 (2000).LSMEDI0196-8092 http://dx.doi.org/10.1002/(ISSN)1096-9101 Google Scholar

41. Y. Liu et al., “Investigation of depth selectivity of polarization gating for tissue characterization,” Opt. Express 13(2), 601–611 (2005).OPEXFF1094-4087 http://dx.doi.org/10.1364/OPEX.13.000601 Google Scholar

42. S. G. Demos and R. R. Alfano, “Temporal gating in highly scattering media by the degree of optical polarization,” Opt. Lett. 21(2), 161–163 (1996).OPLEDP0146-9592 http://dx.doi.org/10.1364/OL.21.000161 Google Scholar

43. S. P. Morgan and M. E. Ridgway, “Polarization properties of light backscattered from a two layer scattering medium,” Opt. Express 7(12), 395–402 (2000).OPEXFF1094-4087 http://dx.doi.org/10.1364/OE.7.000395 Google Scholar

44. S. McGee et al., “Anatomy-based algorithms for detecting oral cancer using reflectance and fluorescence spectroscopy,” Ann. Otol. Rhinol. Laryngol. 119(11), 817–826 (2010). http://dx.doi.org/10.1177/000348941011901112 Google Scholar

45. L. T. Nieman, M. Jakovljevic and K. Sokolov, “Compact beveled fiber optic probe design for enhanced depth discrimination in epithelial tissues,” Opt. Express 17(4), 2780–2796 (2009).OPEXFF1094-4087 http://dx.doi.org/10.1364/OE.17.002780 Google Scholar

46. V. M. Turzhitsky et al., “Measuring mucosal blood supply in vivo with a polarization-gating probe,” Appl. Opt. 47(32), 6046–6057 (2008).APOPAI0003-6935 http://dx.doi.org/10.1364/AO.47.006046 Google Scholar

47. Y. L. Kim et al., “Simultaneous measurement of angular and spectral properties of light scattering for characterization of tissue microarchitecture and its alteration in early precancer,” IEEE J. Sel. Top. Quantum Electron. 9(2), 243–256 (2003). http://dx.doi.org/10.1109/JSTQE.2003.814183 Google Scholar

48. A. Wang, V. Nammalavar and R. Drezek, “Targeting spectral signatures of progressively dysplastic stratified epithelia using angularly variable fiber geometry in reflectance Monte Carlo simulations,” J. Biomed. Opt. 12(4), 044012 (2007).JBOPFO1083-3668 http://dx.doi.org/10.1117/1.2769328 Google Scholar

49. N. Rajaram et al., “Pilot clinical study for quantitative spectral diagnosis of non-melanoma skin cancer,” Lasers Surg. Med. 42(10), 876–887 (2010).LSMEDI0196-8092 http://dx.doi.org/10.1002/lsm.21009 Google Scholar

50. R. Díaz-Uriarte and S. A. De Andres, “Gene selection and classification of microarray data using random forest,” BMC Bioinf. 7(1), 1 (2006).BBMIC41471-2105 http://dx.doi.org/10.1186/1471-2105-7-1 Google Scholar

51. N. Ancona et al., “On the statistical assessment of classifiers using DNA microarray data,” BMC Bioinf. 7(1), 387 (2006).BBMIC41471-2105 http://dx.doi.org/10.1186/1471-2105-7-387 Google Scholar

52. H. J. Van Staveren et al., “Light scattering in lntralipid-10% in the wavelength range of 400–1100 nm,” Appl. Opt. 30(31), 4507–4514 (1991).APOPAI0003-6935 http://dx.doi.org/10.1364/AO.30.004507 Google Scholar

53. S. T. Flock et al., “Optical properties of Intralipid: a phantom medium for light propagation studies,” Lasers Surg. Med. 12(5), 510–519 (1992).LSMEDI0196-8092 http://dx.doi.org/10.1002/(ISSN)1096-9101 Google Scholar

54. P. M. Speight, P. M. Farthing and J. E. Bouquot, “The pathology of oral cancer and precancer,” Current Diagnostic Pathology 3(3), 165–176 (1996). http://dx.doi.org/10.1016/S0968-6053(05)80014-6 Google Scholar

55. R. S. Cotran, V. Kumar and T. Collins, Pathologic Basis of Disease, 6th ed., Saunders, Philadelphia (1999). Google Scholar

56. O. C. Marina, C. K. Sanders and J. R. Mourant, “Effects of acetic acid on light scattering from cells,” J. Biomed. Opt. 17(8), 085002 (2012).JBOPFO1083-3668 http://dx.doi.org/10.1117/1.JBO.17.8.085002 Google Scholar

57. K. Bhalang et al., “The application of acetic acid in the detection of oral squamous cell carcinoma,” Oral Surg. Oral Med. Oral Pathol. Oral Radiol. Endodontol. 106(3), 371–376 (2008). http://dx.doi.org/10.1016/j.tripleo.2008.01.017 Google Scholar

58. L. L. Patton, J. B. Epstein and A. R. Kerr, “Adjunctive techniques for oral cancer examination and lesion diagnosis: a systematic review of the literature,” J. Am. Dent. Assoc. 139(7), 896–905 (2008).JADSAY http://dx.doi.org/10.14219/jada.archive.2008.0276 Google Scholar

59. L. McIntosh, M. J. McCullough and C. S. Farah, “The assessment of diffused light illumination and acetic acid rinse (Microlux/DLTM) in the visualisation of oral mucosal lesions,” Oral Oncol. 45, e227–e231 (2009).EJCCER1368-8375 http://dx.doi.org/10.1016/j.oraloncology.2009.08.001 Google Scholar

60. A. Amelink et al., “Non-invasive measurement of the microvascular properties of non-dysplastic and dysplastic oral leukoplakias by use of optical spectroscopy,” Oral Oncol. 47(12), 1165–1170 (2011).EJCCER1368-8375 http://dx.doi.org/10.1016/j.oraloncology.2011.08.014 Google Scholar

61. L. Qiu et al., “Multispectral scanning during endoscopy guides biopsy of dysplasia in Barrett’s esophagus,” Nat. Med. 16(5), 603–606 (2010).1078-8956 http://dx.doi.org/10.1038/nm.2138 Google Scholar

62. H. R. Roy et al., “Association between rectal optical signatures and colonic neoplasia: potential applications for screening,” Cancer Res. 69(10), 4476–4483 (2009). http://dx.doi.org/10.1158/0008-5472.CAN-08-4780 Google Scholar

63. A. J. Radosevich et al., “Rectal optical markers for in vivo risk stratification of premalignant colorectal lesions,” Clin. Cancer Res. 21(19), 4347–4355 (2015). http://dx.doi.org/10.1158/1078-0432.CCR-15-0136 Google Scholar

64. H. S. Yazdi et al., “Mapping breast cancer blood flow index, composition, and metabolism in a human subject using combined diffuse optical spectroscopic imaging and diffuse correlation spectroscopy,” J. Biomed. Opt. 22(4), 045003 (2017). http://dx.doi.org/10.1117/1.JBO.22.4.045003 Google Scholar

65. J. M. Cochran et al., “Longitudinal optical monitoring of blood flow in breast tumors during neoadjuvant chemotherapy,” Phys. Med. Biol. 62(12), 4637–4653 (2017). Google Scholar

66. S. McGee et al., “Model-based spectroscopic analysis of the oral cavity: impact of anatomy,” J. Biomed. Opt. 13(6), 064034 (2008).JBOPFO1083-3668 http://dx.doi.org/10.1117/1.2992139 Google Scholar

67. A. Nath et al., “Effect of probe pressure on cervical fluorescence spectroscopy measurements,” J. Biomed. Opt. 9(3), 523–533 (2004).JBOPFO1083-3668 http://dx.doi.org/10.1117/1.1695562 Google Scholar

68. S. Ruderman et al., “Analysis of pressure, angle and temporal effects on tissue optical properties from polarization-gated spectroscopic probe measurements,” Biomed. Opt. Express 1(2), 489–499 (2010).BOEICL2156-7085 http://dx.doi.org/10.1364/BOE.1.000489 Google Scholar

69. L. Perelman et al., “Observation of periodic fine structure in reflectance from biological tissue: a new technique for measuring nuclear size distribution,” Phys. Rev. Lett. 80(3), 627–630 (1998).PRLTAO0031-9007 http://dx.doi.org/10.1103/PhysRevLett.80.627 Google Scholar

70. G. Zonios et al., “Diffuse reflectance spectroscopy of human adenomatous colon polyps in vivo,” Appl. Opt. 38(31), 6628–6637 (1999).APOPAI0003-6935 http://dx.doi.org/10.1364/AO.38.006628 Google Scholar

71. D. Arifler et al., “Light scattering from collagen fiber networks: micro-optical properties of normal and neoplastic stroma,” Biophys. J. 92(9), 3260–3274 (2007).BIOJAU0006-3495 http://dx.doi.org/10.1529/biophysj.106.089839 Google Scholar

72. F. E. Robles et al., “Molecular imaging true-colour spectroscopic optical coherence tomography,” Nat. Photonics 5(12), 744–747 (2011).NPAHBY1749-4885 http://dx.doi.org/10.1038/nphoton.2011.257 Google Scholar

73. F. Robles, R. N. Graf and A. Wax, “Dual window method for processing spectroscopic optical coherence tomography signals with simultaneously high spectral and temporal resolution,” Opt. Express 17(8), 6799–6812 (2009).OPEXFF1094-4087 http://dx.doi.org/10.1364/OE.17.006799 Google Scholar

74. M. J. Bailey and K. Sokolov, “Depth-resolved measurements with elliptically polarized reflectance spectroscopy,” Biomed. Opt. Express 7(7), 2861–2876 (2016).BOEICL2156-7085 http://dx.doi.org/10.1364/BOE.7.002861 Google Scholar

75. A. Da Silva, C. Deumié and I. Vanzetta, “Elliptically polarized light for depth resolved optical imaging,” Biomed. Opt. Express 3(11), 2907–2915 (2012).BOEICL2156-7085 http://dx.doi.org/10.1364/BOE.3.002907 Google Scholar

76. S. Rehn et al., “Depth probing of diffuse tissues controlled with elliptically polarized light,” J. Biomed. Opt. 18(1), 016007 (2013).JBOPFO1083-3668 http://dx.doi.org/10.1117/1.JBO.18.1.016007 Google Scholar

Biographies for the authors are not available.

© The Authors. 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.
Maria J. Bailey, Nishant Verma, Leonid Fradkin, Sylvia Lam, Calum MacAulay, Catherine Poh, Mia K. Markey, Konstantin Sokolov, "Detection of precancerous lesions in the oral cavity using oblique polarized reflectance spectroscopy: a clinical feasibility study," Journal of Biomedical Optics 22(6), 065002 (13 June 2017). http://dx.doi.org/10.1117/1.JBO.22.6.065002 Submission: Received 6 March 2017; Accepted 18 May 2017
Submission: Received 6 March 2017; Accepted 18 May 2017
JOURNAL ARTICLE
19 PAGES


SHARE
KEYWORDS
Reflectance spectroscopy

Spectroscopy

Binary data

Diagnostics

In vivo imaging

Optical fibers

Polarization

RELATED CONTENT

Shaping of dark beams in resonators with a bi lens...
Proceedings of SPIE (February 17 2010)
An Optical Fiber Flowmeter Based On The Coriolis Effect
Proceedings of SPIE (February 01 1989)
Digital Disk Optics
Proceedings of SPIE (November 10 1983)

Back to Top