Fiber-bundle microendoscopy with sub-diffuse reflectance spectroscopy and intensity mapping for multimodal optical biopsy of stratified epithelium

Sub-diffuse Reflectance Spectroscopy Combined With Machine Learning Method for Oral Mucosal Disease Identification

OBJECTIVES

Oral squamous cell carcinoma (OSCC) is the sixth-highest incidence of malignant tumors worldwide. However, early diagnosis is complex owing to the impracticality of biopsying every potentially premalignant intraoral lesion. Here, we present a sub-diffuse reflectance spectroscopy combined with a machine learning method for oral mucosal disease identification. This method provides a noninvasive cost-effective identification option for early signs of malignancy.

METHODS

Sub-diffuse spectra of three oral sites (hypoglottis, buccal, and gingiva) from healthy subjects and three types of mucosal lesions (oral lichen planus, OLP, oral leukoplakia, OLK, and OSCC) from patients were collected by using a home-made sub-diffuse reflectance spectroscopy prototype system, and three features including spectra ratio (SR), first-order derivative(DE) of the spectra and optical parameters (OP) were derived from the original spectra to enhance the insights into the optical properties of the oral mucosal tissues. To accurately classify the spectral features, a support vector machine (SVM) and probabilistic neural network (PNN) were employed.

RESULT

Most of the statistical distributions of the spectral features demonstrated obvious differences and the two classification methods exhibited comparable performances. For the classification in the oral sites of healthy subjects, the OP-based classification results were unsatisfactory, while the classification results utilizing DR, SR, and DE achieved a least accuracy of 0.8289, sensitivity of 0.8495, sensitivity of 0.9311, and Matthews correlation coefficient of 0.8085. Comparatively, the classification results between OLP, OLK, OSCC, and normal tissue obtained achieved high indexes even using the OP feature.

CONCLUSION

Integrating sub-diffuse reflectance spectroscopy measurement and suitable machine learning methods can obtain remarkable precision in differentiating different sites of oral mucosa and identifying different types of oral mucosal diseases, especially based on DE features. It is of great help in detecting OSCC and is expected to be a highly sensitive, time-sensitive, and accurate method for oral disease detection.

Neural network-based optimization of sub-diffuse reflectance spectroscopy for improved parameter prediction and efficient data collection

In this study, a general and systematical investigation of sub-diffuse reflectance spectroscopy is implemented. A Gegenbauer-kernel phase function-based Monte Carlo is adopted to describe photon transport more efficiently. To improve the computational efficiency and accuracy, two neural network algorithms, namely, back propagation neural network and radial basis function neural network are utilized to predict the absorption coefficient μ a , reduced scattering coefficient μ s ' and sub-diffusive quantifier γ , simultaneously, at multiple source-detector separations (SDS). The predicted results show that the three parameters can be predicated accurately by selecting five SDSs or above. Based on the simulation results, a four wavelength (520, 650, 785 and 830 nm) measurement system using five SDSs is designed by adopting phase-lock-in technique. Furtherly, the trained neural-network models are utilized to extract optical properties from the phantom and in vivo experimental data. The results verify the feasibility and effectiveness of our proposed system and methods in mucosal disease diagnosis.

Macrophage-targeted anti-CCL2 immunotherapy enhances tumor sensitivity to 5-fluorouracil in a Balb/c-CT26 murine colon carcinoma model measured using diffuse reflectance spectroscopy

Background

Immunotherapy in colorectal cancer (CRC) regulates specific immune checkpoints and, when used in combination with chemotherapy, can improve patient prognosis. One specific immune checkpoint is the recruitment of circulating monocytes that differentiate into tumor-associated macrophages (TAMs) and promote tumor angiogenesis. Changes in vascularization can be non-invasively assessed via diffuse reflectance spectroscopy using hemoglobin concentrations and oxygenation in a localized tumor volume. In this study, we examine whether blockade of monocyte recruitment via CCL2 (macrophage chemoattractant protein-1) leads to enhanced sensitivity of 5-fluorouracil (5-FU) in a CT26-Balb/c mouse model of CRC. It was hypothesized that the blockade of TAMs will alter tumor perfusion, increasing chemotherapy response. A subcutaneous tumor model using Balb/c mice injected with CT26 colon carcinoma cells received either a saline or isotype control, anti-CCL2, 5-FU, or a combination of anti-CCL2 and 5-FU.

Results

Findings show that 12 days post-treatment, monocyte recruitment was significantly reduced by approximately 61% in the combination group. This shows that the addition of anti-CCL2 to 5-FU slowed the fold-change (change from the original measurement to the final measurement) in tumor volume from Day 0 to Day 12 (~ 5 fold). Modest improvements in oxygen saturation (~ 30%) were observed in the combination group.

Conclusion

The findings in this work suggest that the blockade of CCL2 is sufficient in the reduction of TAMs that are recruited into the tumor microenvironment and has the ability to modestly alter tumor perfusion during early-tumor response to treatment even though the overall benefit is relatively modest.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12865-022-00493-5.

Application of Classification Algorithms to Diffuse Reflectance Spectroscopy Measurements for Ex Vivo Characterization of Biological Tissues

Biological tissue identification in real clinical scenarios is a relevant and unsolved medical problem, particularly in the operating room. Although it could be thought that healthy tissue identification is an immediate task, in practice there are several clinical situations that greatly impede this process. For instance, it could be challenging in open surgery in complex areas, such as the neck, where different structures are quite close together, with bleeding and other artifacts affecting visual inspection. Solving this issue requires, on one hand, a high contrast noninvasive technique and, on the other hand, powerful classification algorithms. Regarding the technique, optical diffuse reflectance spectroscopy has demonstrated such capabilities in the discrimination of tumoral and healthy biological tissues. The complex signals obtained, in the form of spectra, need to be adequately computed in order to extract relevant information for discrimination. As usual, accurate discrimination relies on massive measurements, some of which serve as training sets for the classification algorithms. In this work, diffuse reflectance spectroscopy is proposed, implemented, and tested as a potential technique for healthy tissue discrimination. A specific setup is built and spectral measurements on several ex vivo porcine tissues are obtained. The massive data obtained are then analyzed for classification purposes. First of all, considerations about normalization, detrending and noise are taken into account. Dimensionality reduction and tendencies extraction are also considered. Featured spectral characteristics, principal component or linear discrimination analysis are applied, as long as classification approaches based on k-nearest neighbors (k-NN), quadratic discrimination analysis (QDA) or Naïve Bayes (NB). Relevant parameters about classification accuracy are obtained and compared, including ANOVA tests. The results show promising values of specificity and sensitivity of the technique for some classification algorithms, even over 95%, which could be relevant for clinical applications in the operating room.

Simple analytical total diffuse reflectance over a reduced-scattering-pathlength scaled dimension of [10-5, 10-1] from a medium with HG scattering anisotropy

Model approximation is necessary for reflectance assessment of tissue at sub-diffusive to non-diffusive scale. For tissue probing over a sub-diffusive circular area centered on the point of incidence, we demonstrate simple analytical steady-state total diffuse reflectance from a semi-infinite medium with the Henyey-Greenstein (HG) scattering anisotropy (factor $g$g). Two physical constraints are abided to: (1) the total diffuse reflectance is the integration of the radial diffuse reflectance; (2) the radial and total diffuse reflectance at $g \gt {0}$g>0 analytically must resort to their respective forms corresponding to isotropic scattering as $g$g becomes zero. Steady-state radial diffuse reflectance near the point of incidence from a semi-infinite medium of $g \approx 0$g≈0 is developed based on the radiative transfer for isotropic scattering, then integrated to find the total diffuse reflectance for $g \approx 0$g≈0. The radial diffuse reflectance for $g \ge 0.5$g≥0.5 is semi-empirically formulated by comparing to Monte Carlo simulation results and abiding to the second constraint. Its integration leads to a total diffuse reflectance for $g \ge 0.5$g≥0.5 that is also bounded by the second constraint. Over a collection diameter of the reduced-scattering pathlength ($1/\mu _s^{ \prime}$1/μs') scaled size of [${{10}^{ - 5}}$10^-5, ${{10}^{ - 1}}$10^-1] for $g = [{0.5},{0.95}]$g=[0.5,0.95] and the absorption coefficient as strong as the reduced scattering coefficient, the simple analytical total diffuse reflectance is found to be accurate, with an average error of 16.1%.

Simple empirical master-slave dual-source configuration within the diffusion approximation enhances modeling of spatially resolved diffuse reflectance at short-path and with low-scattering from a semi-infinite homogeneous medium: erratum

We correct one typographical error of three equations in Appl. Opt.56, 1447 (2017)APOPAI0003-693510.1364/AO.56.001447.

Sampling depth of a diffuse reflectance spectroscopy probe for in-vivo physiological quantification of murine subcutaneous tumor allografts

Diffuse reflectance spectroscopy (DRS) is a probe-based spectral biopsy technique used in cancer studies to quantify tissue reduced scattering (μs') and absorption (μa) coefficients and vary in source-detector separation (SDS) to fine-tune sampling depth. In subcutaneous murine tumor allografts or xenografts, a key design requirement is ensuring that the source light interrogates past the skin layer into the tumor without significantly sacrificing signal-to-noise ratio (target of ≥15  dB). To resolve this requirement, a DRS probe was designed with four SDSs (0.75, 2.00, 3.00, and 4.00 mm) to interrogate increasing tissue volumes between 450 and 900 nm. The goal was to quantify percent errors in extracting μa and μs', and to quantify sampling depth into subcutaneous Balb/c-CT26 colon tumor allografts. Using an optical phantom-based experimental method, lookup-tables were constructed relating μa,μs', diffuse reflectance, and sampling depth. Percent errors were <10  %   and 5% for extracting μa and μs', respectively, for all SDSs. Sampling depth reached up to 1.6 mm at the first Q-band of hemoglobin at 542 nm, the key spectral region for quantifying tissue oxyhemoglobin concentration. This work shows that the DRS probe can accurately extract optical properties and the resultant physiological parameters such as total hemoglobin concentration and tissue oxygen saturation, from sufficient depth within subcutaneous Balb/c-CT26 colon tumor allografts. Methods described here can be generalized for other murine tumor models. Future work will explore the feasibility of the DRS in quantifying volumetric tumor perfusion in response to anticancer therapies.

Effects of isoflurane anesthesia on physiological parameters in murine subcutaneous tumor allografts measured via diffuse reflectance spectroscopy

Diffuse reflectance spectroscopy (DRS) has been used in murine studies to quantify tumor perfusion and therapeutic response. These studies frequently use inhaled isoflurane anesthesia, which depresses the respiration rate and results in the desaturation of arterial oxygen saturation, potentially affecting tissue physiological parameters. However, there have been no controlled studies quantifying the effect of isoflurane anesthesia on DRS-derived physiological parameters of murine tissue. The goal of this study was to perform DRS on Balb/c mouse (n = 10) tissue under various anesthesia conditions to quantify effects on tissue physiological parameters, including total hemoglobin concentration, tissue oxygen saturation, oxyhemoglobin and reduced scattering coefficient. Two independent variables were manipulated including metabolic gas type (pure oxygen vs. medical air) and isoflurane concentration (1.5 to 4.0%). The 1.5% isoflurane and 1 L/min oxygen condition most closely mimicked a no-anesthesia condition with oxyhemoglobin concentration within 89% ± 19% of control. The time-dependent effects of isoflurane anesthesia were tested, revealing that anesthetic induction with 4.0% isoflurane can affect DRS-derived physiological parameters up to 20 minutes post-induction. Finally, spectroscopy with and without isoflurane anesthesia was compared for colon tumor Balb/c-CT26 allografts (n = 5) as a representative model of subcutaneous murine tumor allografts. Overall, isoflurane anesthesia yielded experimentally-induced depressed oxyhemoglobin, and this depression was both concentration and time dependent. Investigators should understand the dynamic effects of isoflurane on tissue physiological parameters measured by DRS. These results may guide investigators in eliminating, limiting, or managing anesthesia-induced physiological changes in DRS studies in mouse models.

Multimodal Imaging and Spectroscopy Fiber-bundle Microendoscopy Platform for Non-invasive, In Vivo Tissue Analysis

Recent fiber-bundle microendoscopy techniques enable non-invasive analysis of in vivo tissue using either imaging techniques or a combination of spectroscopy techniques. Combining imaging and spectroscopy techniques into a single optical probe may provide a more complete analysis of tissue health. In this article, two dissimilar modalities are combined, high-resolution fluorescence microendoscopy imaging and diffuse reflectance spectroscopy, into a single optical probe. High-resolution fluorescence microendoscopy imaging is a technique used to visualize apical tissue micro-architecture and, although mostly a qualitative technique, has demonstrated effective real-time differentiation between neoplastic and non-neoplastic tissue. Diffuse reflectance spectroscopy is a technique which can extract tissue physiological parameters including local hemoglobin concentration, melanin concentration, and oxygen saturation. This article describes the specifications required to construct the fiber-optic probe, how to build the instrumentation, and then demonstrates the technique on in vivo human skin. This work revealed that tissue micro-architecture, specifically apical skin keratinocytes, can be co-registered with its associated physiological parameters. The instrumentation and fiber-bundle probe presented here can be optimized as either a handheld or endoscopically-compatible device for use in a variety of organ systems. Additional clinical research is needed to test the viability of this technique for different epithelial disease states.

Diffuse optical microscopy for quantification of depth-dependent epithelial backscattering in the cervix

A fiber optic imaging approach is presented using structured illumination for quantification of almost pure epithelial backscattering. We employ multiple spatially modulated projection patterns and camera-based reflectance capture to image depth-dependent epithelial scattering. The potential diagnostic value of our approach is investigated on cervical ex vivo tissue specimens. Our study indicates a strong backscattering increase in the upper part of the cervical epithelium caused by dysplastic microstructural changes. Quantization of relative depth-dependent backscattering is confirmed as a potentially useful diagnostic feature for detection of precancerous lesions in cervical squamous epithelium.

Towards monitoring dysplastic progression in the oral cavity using a hybrid fiber-bundle imaging and spectroscopy probe

Intraepithelial dysplasia of the oral mucosa typically originates in the proliferative cell layer at the basement membrane and extends to the upper epithelial layers as the disease progresses. Detection of malignancies typically occurs upon visual inspection by non-specialists at a late-stage. In this manuscript, we validate a quantitative hybrid imaging and spectroscopy microendoscope to monitor dysplastic progression within the oral cavity microenvironment in a phantom and pre-clinical study. We use an empirical model to quantify optical properties and sampling depth from sub-diffuse reflectance spectra (450–750 nm) at two source-detector separations (374 and 730 μm). Average errors in recovering reduced scattering (5–26 cm⁻¹) and absorption coefficients (0–10 cm⁻¹) in hemoglobin-based phantoms were approximately 2% and 6%, respectively. Next, a 300 μm-thick phantom tumor model was used to validate the probe’s ability to monitor progression of a proliferating optical heterogeneity. Finally, the technique was demonstrated on 13 healthy volunteers and volume-averaged optical coefficients, scattering exponent, hemoglobin concentration, oxygen saturation, and sampling depth are presented alongside a high-resolution microendoscopy image of oral mucosa from one volunteer. This multimodal microendoscopy approach encompasses both structural and spectroscopic reporters of perfusion within the tissue microenvironment and can potentially be used to monitor tumor response to therapy.

In vivo measurement of non-keratinized squamous epithelium using a spectroscopic microendoscope with multiple source-detector separations

In the non-keratinized epithelia, dysplasia typically arises near the basement membrane and proliferates into the upper epithelial layers over time. We present a non-invasive, multimodal technique combining high-resolution fluorescence imaging and broadband sub-diffuse reflectance spectroscopy (sDRS) to monitor health at various tissue layers. This manuscript focuses on characterization of the sDRS modality, which contains two source-detector separations (SDSs) of 374 μm and 730 μm, so that it can be used to extract in vivo optical parameters from human oral mucosa at two tissue thicknesses. First, we present empirical lookup tables (LUTs) describing the relationship between reduced scattering (μ_s') and absorption coefficients (μ_a) and absolute reflectance. LUTS were shown to extract μ_s' and μ_a with accuracies of approximately 4% and 8%, respectively. We then present LUTs describing the relationship between μ_s', μ_a and sampling depth. Sampling depths range between 210-480 and 260-620 μm for the 374 and 730 μm SDSs, respectively. We then demonstrate the ability to extract in vivo μ_s', μ_a, hemoglobin concentration, bulk tissue oxygen saturation, scattering exponent, and sampling depth from the inner lip of thirteen healthy volunteers to elucidate the differences in the extracted optical parameters from each SDS (374 and 730 μm) within non-keratinized squamous epithelia.