Determination of optical parameters of human breast tissue from spatially resolved fluorescence: a diffusion theory model

Spatially Resolved Fibre-Optic Probe for Cervical Precancer Detection Using Fluorescence Spectroscopy and PCA-ANN-Based Classification Algorithm: An In Vitro Study

Cervical cancer can be detected at an early stage through the changes occurring in biochemical and morphological properties of epithelium layer. Fluorescence spectroscopy has the ability to identify these subtle changes non-invasively and in real time with good accuracy in comparison with conventional techniques. In this paper, we report the usage of a custom designed spatially resolved fibre-optic probe (SRFOP), which consists of 77 fibres in two concentric rings, for the detection of cervical cancer using fluorescence spectroscopy technique. The aim of this study is to classify different grades of cervical precancer on the basis of their fluorescence spectra followed by a robust classification algorithm. Fluorescence spectra of 28 cervical tissue samples of different categories have been recorded using six detector fibres of FOP at different spatial locations with the source fibre (SF). A 405 nm laser diode source has been utilised to excite the samples and a USB 4000 Ocean Optics spectrometer to collect the output spectra in the wavelength range 400-700 nm. Principal component analysis (PCA) was applied to the collected spectra to reduce the dimensionality of the data while preserving the most significant features for classification. The first 10 principal components, which captured the majority of the variance in the spectra, were selected as input features for the classification model. Classification was then performed using an artificial neural network (ANN) with a specific architecture, including an input layer, hidden layers, and a softmax activation function in the output layer. Experimental and classification results both demonstrate that proximal fibres (PFs) perform better than distal fibres (DFs) in capturing the discriminatory features present in the epithelium layer of cervical tissue samples as PF collect most of the signal from the epithelium layer. The combined approach of spatially resolved fluorescence spectroscopy and PCA-ANN classification techniques is able to discriminate different grades of cervical precancer and normal with an average sensitivity, specificity and accuracy of 93.33%, 96.67% and 95.57%, respectively.

Wavelet scattering transform and entropy features in fluorescence spectral signal analysis for cervical cancer diagnosis

Cervical cancer is a prevalent malignant tumor within the female reproductive system and is regarded as a prominent cause of female mortality on a global scale. Timely and precise detection of various phases of cervical cancer holds the potential to substantially enhance both the rate of successful treatment and the duration of patient survival. Fluorescence spectroscopy is a highly sensitive method for detecting the biochemical changes that arise during cancer progression. In our study, fluorescence spectral data is collected from a diverse group of 110 subjects. The potential of the scattering transform technique for the purpose of cancer detection is explored. The processed signal undergoes an initial decomposition into scattering coefficients using the wavelet scattering transform (WST). Subsequently, the scattering coefficients are subjected to computation for fuzzy entropy, dispersion entropy, phase entropy, and spectral entropy, for effectively characterizing the fluorescence spectral signals. These combined features generated through the proposed approach are then fed to 1D convolutional neural network (CNN) classifier to classify them into normal, pre-cancerous, and cancerous categories, thereby evaluating the effectiveness of the proposed methodology. We obtained mean classification accuracy of 97% using 5-fold cross-validation. This demonstrates the potential of combining WST and entropic features for analyzing fluorescence spectroscopy signals using 1D CNN classifier that enables early cancer detection in contrast to prevailing diagnostic methods.

Cervical pre-cancer classification using entropic features and CNN: In vivo validation with a handheld fluorescence probe

Cervical cancer is one of the most prevalent forms of cancer, with a lengthy latent period and a gradual onset phase. Conventional techniques are found to be severely lacking in real time detection of disease progression which can greatly enhance the cure rate. Due to their high sensitivity and specificity, optical techniques are emerging as reliable tools, particularly in case of cancer. It has been seen that biochemical changes are better highlighted through intrinsic fluorescence devoid of interference from absorption and scattering. Its effectiveness in in-vivo conditions is affected by the fact that the intrinsic spectral signatures vary from patient to patient, as well as in different population groups. Here, we overcome this limitation by collectively enumerating the subtle changes in the spectral profiles and correlations through an information theory based entropic approach, which significantly amplifies the minute spectral variations. In conjunction with artificial intelligence (AI)/machine learning (ML) tools, it yields high specificity and sensitivity with a small dataset from patients in clinical conditions, without artificial augmentation. We have used an in-house developed handheld probe (i-HHP) for extracting intrinsic fluorescence spectra of human cervix from 110 different subjects drawn from diverse population groups. The average classification accuracy of the proposed methodology using 10-fold cross validation is 93.17%. A combination of polarised fluorescence spectra from i-HHP and the proposed classifier is proven to be minimally invasive with the ability to diagnose patients in real time. This paves the way for effective use of relatively smaller sized sensitive fluorescence data with advanced AI/ML tools for early cervical cancer detection in clinics.

Optical property recovery with spatially-resolved diffuse reflectance at short source-detector separations using a compact fiber-optic probe

We describe a compact fiber-optic probe (2 mm outside diameter) that utilizes spatially-resolved diffuse reflectance for tissue optical property recovery. Validation was performed in phantoms containing Intralipid 20% as scatterer, and methylene blue (MB), MnTPPS, and/or India ink as absorbers. Over a range of conditions, the reduced scattering coefficient was recovered with a root mean square error (RMSE) of 0.86-2.7 cm^-1 (average error = 3.8%). MB concentration was recovered with RMSE = 0.26-0.52 µM (average error = 15.0%), which did not vary with inclusion of MnTPPS (p=0.65). This system will be utilized to determine optical properties in human abscesses, in order to generate treatment plans for photodynamic therapy.

Concentration of FAD as a marker for cervical precancer detection

We report the ex vivo results of an in-house fabricated portable device based on polarized fluorescence measurements in the clinical environment. This device measures the polarized fluorescence and elastic scattering spectra with 405-nm laser and white light sources, respectively. The dominating fluorophore with 405-nm excitation is flavin adenine dinucleotide (FAD) with a fluorescence peak around 510 nm. The measured spectra are highly modulated by the interplay of scattering and absorption effects. Due to this, valuable information gets masked. To reduce these effects, intrinsic fluorescence was extracted by normalizing polarized fluorescence spectra with polarized elastic scattering spectra obtained. A number of fluorophores contribute to the fluorescence spectra and need to be decoupled to understand their roles in the progression of cancer. Nelder-Mead method has been utilized to fit the spectral profile with Gaussian to decouple the different bands of contributing fluorophores (FAD and porphyrin). The change in concentration of FAD during disease progression manifests in the change in ratio of total area to FWHM of its Gaussian profile. Receiver operating characteristic (ROC) curve analysis has been used to discriminate different grades of cervical precancer by using the ratio as input parameter. The sensitivity and specificity for discrimination of normal samples from CIN I (cervical intraepithelial neoplasia) are 75% and 54%, respectively. Further, the normal samples can be discriminated from CIN II samples with 100% and 82% sensitivity and specificity, respectively, and the CIN I from CIN II samples can also be discriminated with 100% sensitivity and 90% specificity, respectively. The results show that the change in the concentration of (FAD) can be used as a marker to discriminate the different grades of the cancer and biochemical changes at an early stage of the cancer can also be monitored with this technique.

Detecting cervical cancer progression through extracted intrinsic fluorescence and principal component analysis

Intrinsic fluorescence spectra of the human normal, cervical intraepithelial neoplasia 1 (CIN1), CIN2, and cervical cancer tissue have been extracted by effectively combining the measured polarized fluorescence and polarized elastic scattering spectra. The efficacy of principal component analysis (PCA) to disentangle the collective behavior from smaller correlated clusters in a dimensionally reduced space in conjunction with the intrinsic fluorescence is examined. This combination unambiguously reveals the biochemical changes occurring with the progression of the disease. The differing activities of the dominant fluorophores, collagen, nicotinamide adenine dinucleotide, flavins, and porphyrin of different grades of precancers are clearly identified through a careful examination of the sectorial behavior of the dominant eigenvectors of PCA. To further classify the different grades, the Mahalanobis distance has been calculated using the scores of selected principal components.

Single-molecule super-resolution light-sheet microscopy

Single-molecule super-resolution imaging is a new promising tool for investigation of sub-cellular structures. Concurrently, light-sheet microscopy, also known as selective plane illumination microscopy (SPIM), has gained rapid favor with the imaging community in developmental biology due to its fast speed, high contrast, deep penetration, and low phototoxicity. While nearly a dozen reviews thoroughly describe the development of light-sheet microscopy and its technological breakthroughs with a main focus on improving the 3D imaging speed of fish embryos, central nervous system, and other tissues, few have addressed the potential of combining light-sheet microscopy and localization-based super-resolution imaging to achieve sub-diffraction-limited resolution. Adapting light-sheet illumination for single-molecule imaging presents unique challenges for instrumentation and reconstruction algorithms. In this Minireview, we provide an overview of the recent developments that address these challenges. We compare different approaches in super-resolution and light-sheet imaging, address advantages and limitations in each approach, and outline future directions of this emerging field.

Characteristic spectral features of the polarized fluorescence of human breast cancer in the wavelet domain

Wavelet transform of polarized fluorescence spectra of human breast tissues is found to localize spectral features that can reliably differentiate normal and malignant tissue types. The intensity differences of parallel and perpendicularly polarized fluorescence spectra are subjected to investigation, since they are relatively free of diffusive background. A number of parameters, capturing spectral variations and subtle changes in the diseased tissues in the visible wavelength regime, are clearly identifiable in the wavelet domain. These manifest both in the average low-pass and high frequency high-pass wavelet coefficients.

A review of in-vivo optical properties of human tissues and its impact on PDT

A thorough understanding of optical properties of biological tissues is critical to effective treatment planning for therapies such as photodynamic therapy (PDT). In the last two decades, new technologies, such as broadband diffuse spectroscopy, have been developed to obtain in vivo data in humans that was not possible before. We found that the in vivo optical properties generally vary in the ranges μ(a) = 0.03-1.6 cm⁻¹ and μ'(s) = 1.2-40 cm⁻¹, although the actual range is tissue-type dependent. We have also examined the overall trend of the absorption spectra (for μ(a) and μ'(s)) as a function of wavelength within a 95% confidence interval for various tissues in vivo. The impact of optical properties on light fluence rate is also discussed for various light application geometries including superficial, interstitial, and within a cavity.

Signal detectability in diffusive media using phased arrays in conjunction with detector arrays

We investigate Hotelling observer performance (i.e., signal detectability) of a phased array system for tasks of detecting small inhomogeneities and distinguishing adjacent abnormalities in uniform diffusive media. Unlike conventional phased array systems where a single detector is located on the interface between two sources, we consider a detector array, such as a CCD, on a phantom exit surface for calculating the Hotelling observer detectability. The signal detectability for adjacent small abnormalities (2 mm displacement) for the CCD-based phased array is related to the resolution of reconstructed images. Simulations show that acquiring high-dimensional data from a detector array in a phased array system dramatically improves the detectability for both tasks when compared to conventional single detector measurements, especially at low modulation frequencies. It is also observed in all studied cases that there exists the modulation frequency optimizing CCD-based phased array systems, where detectability for both tasks is consistently high. These results imply that the CCD-based phased array has the potential to achieve high resolution and signal detectability in tomographic diffusive imaging while operating at a very low modulation frequency. The effect of other configuration parameters, such as a detector pixel size, on the observer performance is also discussed.

Models of light propagation in human tissue applied to cancer diagnostics

Optical methods such as reflectance and fluorescence spectroscopy are being investigated for their potential to aid cancer detection in a quantitative, minimally invasive manner. Mathematical models of reflectance and fluorescence provide an important link between measured optical data and biomedically-relevant tissue parameters that can be extracted from these data to characterize the presence or absence of disease. The most commonly-used mathematical models in biomedical optics are the diffusion approximation (DA) to the radiative transfer equation, Monte Carlo (MC) computational models of light transport, and semi-empirical models. This paper presents a review of the applications of these models to reflectance and endogenous fluorescence sensing for cancer diagnostics in human tissues. Specific examples are given for cervical, breast, and pancreatic tissues. A comparison of the DA and MC methods in two biologically-relevant regimes of optical parameter space will also be discussed.

Diagnosis of vulvar lesions by non-invasive optical analysis: a pilot study

A procedure that could allow an early in vivo and non-invasive detection of vulvar lesions would be extremely useful. We tested an innovative optical method (Optiprobe), which uses a harmless, visible light source for the in vivo, on-line detection of minimal alterations in the structure of vulvar epithelium. A group of 3 female volunteers without gynecological symptoms were first screened to evaluate optical properties of normal vulvar tissue. Next, a group of 16 patients undergoing gynecological examination for vulvar lesions was evaluated by the Optiprobe at suspected sites before these sites were biopsied for histological analysis. Adjacent, non-involved sites were also measured to provide internal controls. Histological analysis of the biopsies identified one case that did not show obvious alterations, 4 cases of high-grade vulvar intraepithelial neoplasia (VIN), 5 cases of vulvitis, and 6 cases of lichen sclerosis (LS).

The optical properties of the VIN cases were significantly different from those of controls, due to a decrease in the absorption spectra and an increase in the scattering spectra. In contrast, a significant increase in the absorption spectra and a decrease in the scattering spectra were observed in the cases of vulvitis. In the LS cases, the absorption spectra were as in controls, whereas the scattering spectra were significantly decreased. We conclude that the Optiprobe provides a useful tool for a rapid and non-invasive detection of vulvar alterations. The method should contribute to reduce the number of biopsies and to facilitate the long-term follow-up of vulvar lesions.

Monte Carlo simulation of time-dependent, transport-limited fluorescent boundary measurements in frequency domain

Recently, we have presented and experimentally validated a unique numerical solver of the coupled radiative transfer equations (RTEs) for rapidly computing time-dependent excitation and fluorescent light propagation in small animal tomography. Herein, we present a time-dependent Monte Carlo algorithm to validate the forward RTE solver and investigate the impact of physical parameters upon transport-limited measurements in order to best direct the development of the RTE solver for optical tomography. Experimentally, the Monte Carlo simulations for both transport-limited and diffusion-limited propagations are validated using frequency domain photon migration measurements for 1.0%, 0.5%, and 0.2% intralipid solutions containing 1 microM indocyanine green in a 49 cm3 cylindrical phantom corresponding to the small volume employed in small animal tomography. The comparisons between Monte Carlo simulations and the numerical solutions result in mean percent error in amplitude and the phase shift less than 5.0% and 0.7 degrees, respectively, at excitation and emission wavelengths for varying anisotropic factors, lifetimes, and modulation frequencies. Monte Carlo simulations indicate that the accuracy of the forward model is enhanced using (i) suitable source models of photon delivery, (ii) accurate anisotropic factors, and (iii) accurate acceptance angles of collected photons. Monte Carlo simulations also show that the accuracy of the diffusion approximation in the small phantom depends upon (i) the ratio d(phantom)/l(tr), where d(phantom) is the phantom diameter and l(tr) is the transport mean free path; and (ii) the anisotropic factor of the medium. The Monte Carlo simulations validates and guides the future development of an appropriate RTE solver for deployment in small animal optical tomography.

Simultaneous extraction of optical transport parameters and intrinsic fluorescence of tissue mimicking model media using a spatially resolved fluorescence technique

We present a method based on spatially resolved fluorescence measurement for the simultaneous estimation of optical transport parameters, namely, the reduced scattering coefficient (micro s'), the absorption coefficient (micro a), and the intrinsic fluorescence spectra from turbid media. The accuracy of this approach was tested by conducting studies on a series of tissue-simulating phantoms with known optical transport properties. The estimated relative error in the values for micro s' and micro a using this technique was found to be < or =10%. Furthermore, the line shape and intensity of the intrinsic fluorescence recovered by using this approach were observed to be free from the distorting effects of the wavelength-dependent absorption and scattering properties of the medium, and they were in excellent agreement with the directly measured intrinsic fluorescence spectra of the fluorophores.

Model-based analysis of clinical fluorescence spectroscopy for in vivo detection of cervical intraepithelial dysplasia

We present a mathematical model to calculate the relative concentration of light scatterers, light absorbers, and fluorophores in the epithelium and stroma. This mathematical description is iteratively fit to the fluorescence spectra measured in vivo, yielding relative concentrations of each molecule. The mathematical model is applied to a total of 493 fluorescence measurements of normal and dysplastic cervical tissue acquired in vivo from 292 patients. The estimated parameters are compared with histopathologic diagnosis to evaluate their diagnostic potential. The mathematical model is validated using fluorescence spectra simulated with known sets of optical parameters. Subsequent application of the mathematical model to in vivo fluorescence measurements from cervical tissue yields fits that accurately describe measured data. The optical parameters estimated from 493 fluorescence measurements show an increase in epithelial flavin adenine dinucleotide (FAD) fluorescence, a decrease in epithelial keratin fluorescence, an increase in epithelial light scattering, a decrease in stromal collagen fluorescence, and an increase in stromal hemoglobin light absorption in dysplastic tissue compared to normal tissue. These changes likely reflect an increase in the metabolic activity and loss of differentiation of epithelial dysplastic cells, and stromal angiogenesis associated with dysplasia. The model presented here provides a tool to analyze clinical fluorescence spectra yielding quantitative information about molecular changes related to dysplastic transformation.

Diagnosis and grading of gastritis by non-invasive optical analysis

OBJECTIVES

The precise identification of many diseases of the gastrointestinal tract requires the histological analysis of multiple biopsies of the lining mucosae, thus preventing an immediate diagnosis and the safe screening of the entire organ. To address these limitations, we developed a novel spectroscopic procedure for a real-time, non-invasive optical analysis of mucosae.

METHODS

We have used a fibre-optic probe that monitors light propagation through small tissue volumes to evaluate the antral and fundic mucosa of 51 patients that underwent gastroscopy for symptoms of dyspepsia. Several optical coefficients were computed from the recorded light reflectance, and confronted to the diagnosis made by an expert gastroscopist at the time of the clinical examination. Both evaluations were then validated by comparison with the histological diagnosis of a pathologist who screened biopsies taken at the sites of the optical measurements.

RESULTS

We report that the optical procedure discriminated normal and pathological gastric mucosae with a higher sensitivity and specificity than endoscopic diagnosis. We also show that the changes in light-scattering coefficient, which permitted the optical diagnosis of gastritis alterations, were indirectly correlated with the extent of inflammatory infiltration of the mucosa and detected mucosal alterations mild enough to escape endoscopic detection.

CONCLUSIONS

The results show that, in a normal clinical setting, the optical in vivo analysis provided by our system detects alterations typical of gastritis, and allow for their graded scoring with a specificity and sensitivity that compare well with those of standard histology, while avoiding the invasiveness of the latter procedure. The method is adaptable to the screening of other types of lesions and mucosae and, hence, should prove useful in improving available diagnostic approaches.

Analytical model to describe fluorescence spectra of normal and preneoplastic epithelial tissue: comparison with Monte Carlo simulations and clinical measurements

Fluorescence spectroscopy has shown promise for the detection of precancerous changes in vivo. The epithelial and stromal layers of tissue have very different optical properties; the albedo is relatively low in the epithelium and approaches one in the stroma. As precancer develops, the optical properties of the epithelium and stroma are altered in markedly different ways: epithelial scattering and fluorescence increase, and stromal scattering and fluorescence decrease. We present an analytical model of the fluorescence spectrum of a two-layer medium such as epithelial tissue. Our hypothesis is that accounting for the two different tissue layers will provide increased diagnostic information when used to analyze tissue fluorescence spectra measured in vivo. The Beer-Lambert law is used to describe light propagation in the epithelial layer, while light propagation in the highly scattering stromal layer is described with diffusion theory. Predictions of the analytical model are compared to results from Monte Carlo simulations of light propagation under a range of optical properties reported for normal and precancerous epithelial tissue. In all cases, the mean square error between the Monte Carlo simulations and the analytical model are within 15%. Finally, model predictions are compared to fluorescence spectra of normal and precancerous cervical tissue measured in vivo; the lineshape of fluorescence agrees well in both cases, and the decrease in fluorescence intensity from normal to precancerous tissue is correctly predicted to within 5%. Future work will explore the use of this model to extract information about changes in epithelial and stromal optical properties from clinical measurements and the diagnostic value of these parameters.

Amending of fluorescence sensor signal localization in human skin by matching of the refractive index

Fluorescence diagnostic techniques are notable amongst many other optical methods because they offer high sensitivity and noninvasive measurement of tissue properties. However, a combination of multiple scattering and physical heterogeneity of biological tissues hampers interpretation of the fluorescence measurements. Analyses of the spatial distribution of endogenous and exogenous fluorophores excitation within tissues and their contribution to the detected signal localization are essential for many applications. We have developed a novel Monte Carlo technique that gives a graphical perception of how the excitation and fluorescence detected signal are localized in tissues. Our model takes into account the spatial distribution of fluorophores, the variation of concentrations and quantum yield. We demonstrate that matching the refractive indices of the ambient medium and topical skin layer improves spatial localization of the detected fluorescence signal within the tissues.

Experimental and theoretical investigation of fluorescence photobleaching and recovery in human breast tissue and tissue phantoms

Photobleaching and recovery of 488-nm excited fluorescence from resected human breast tissue samples have been studied. Profiles of photobleaching decay were seen to be faster in cancerous tissue than in those of the normal tissue. The reverse behavior was observed in profiles of recovery after photobleaching. A theoretical model based on one-dimensional diffusion theory has been developed to provide insight into the phenomena of fluorescence during photobleaching and recovery in a multiply scattering medium such as tissue. To understand photobleaching and recovery with the help of this theoretical model, we carried out experiments with model media that were prepared with authentic fluorophores, scatterers, and absorbers. The results of these studies suggest that the fluorescence photobleaching profiles are affected more by the absorption than by the scattering properties of a turbid medium such as tissue. In contrast, the scattering properties of the medium are found to affect the fluorescence recovery profiles to a greater extent. These observations could be related to the observed difference in fluorescence photobleaching and recovery profiles of normal and cancerous breast tissues.

Measurement of fluorophore concentrations and fluorescence quantum yield in tissue-simulating phantoms using three diffusion models of steady-state spatially resolved fluorescence

Steady-state diffusion theory models of fluorescence in tissue have been investigated for recovering fluorophore concentrations and fluorescence quantum yield. Spatially resolved fluorescence, excitation and emission reflectance Carlo simulations, and measured using a multi-fibre probe on tissue-simulating phantoms containing either aluminium phthalocyanine tetrasulfonate (AlPcS4), Photofrin meso-tetra-(4-sulfonatophenyl)-porphine dihydrochloride The accuracy of the fluorophore concentration and fluorescence quantum yield recovered by three different models of spatially resolved fluorescence were compared. The models were based on: (a) weighted difference of the excitation and emission reflectance, (b) fluorescence due to a point excitation source or (c) fluorescence due to a pencil beam excitation source. When literature values for the fluorescence quantum yield were used for each of the fluorophores, the fluorophore absorption coefficient (and hence concentration) at the excitation wavelength (mu(a,x,f)) was recovered with a root-mean-square accuracy of 11.4% using the point source model of fluorescence and 8.0% using the more complicated pencil beam excitation model. The accuracy was calculated over a broad range of optical properties and fluorophore concentrations. The weighted difference of reflectance model performed poorly, with a root-mean-square error in concentration of about 50%. Monte Carlo simulations suggest that there are some situations where the weighted difference of reflectance is as accurate as the other two models, although this was not confirmed experimentally. Estimates of the fluorescence quantum yield in multiple scattering media were also made by determining mu(a,x,f) independently from the fitted absorption spectrum and applying the various diffusion theory models. The fluorescence quantum yields for AlPcS4 and TPPS4 were calculated to be 0.59 +/- 0.03 and 0.121 +/- 0.001 respectively using the point source model, and 0.63 +/- 0.03 and 0.129 +/- 0.002 using the pencil beam excitation model. These results are consistent with published values.