Structurally anisotropic materials are ubiquitous in several application fields, yet their accurate optical characterization remains challenging due to the lack of general models linking their scattering coefficients to the macroscopic transport observables and the need to combine multiple measurements to retrieve their direction-dependent values. Here, we present an improved method for the experimental determination of light-transport tensor coefficients from the diffusive rates measured along all three directions, based on transient transmittance measurements and a generalized Monte Carlo model. We apply our method to the characterization of light-transport properties in two common anisotropic materials—polytetrafluoroethylene tape and paper—highlighting the magnitude of systematic deviations that are typically incurred when neglecting anisotropy.
In many biological tissues such as muscles, dental enamel and mucosa that exhibit macroscopic and/or microscopic spatially anisotropic structures the nature of light scattering becomes anisotropic. This well-known property of tissues is usually neglected, which is rooted in the fact that the available open-source numerical solutions to the radiative transfer equation based on the stochastic Monte Carlo (MC) method do not allow simulations with anisotropic optical properties. In this contribution, we present an extension to our massively parallel PyXOpto (https://github.com/xopto/pyxopto) simulation engine that enables highly efficient and user-friendly MC simulations for layered or voxelated sample geometries with anisotropic scattering properties, both in the steady state and time-resolved domain.
Turbid phantoms play a crucial role in evaluating optical systems and estimating optical properties. Liquid phantoms offer precise tuning of optical properties, but accurately determining their scattering properties is challenging. By using aqueous suspensions of standardized polystyrene microspheres, their optical properties can be theoretically derived using Mie theory. The parameters involved in calculating the scattering coefficient and phase function of microspheres in a liquid medium include the refractive index, density, size probability distribution and solid content of the microspheres and refractive index of the medium. The accuracy of these parameters directly affects the accuracy of calculated optical properties. A lack of clear protocols for phantom preparation and conflicting data in the literature may lead to easily avoidable inaccuracies. We introduce an open-source software that offers a detailed mixing protocol and subsequent optical property calculations for turbid phantoms. The software allows users to input details of the microsphere suspension, target optical property values, and choose between individual or sequentially diluted phantom mixing. It also accommodates the introduction of non-scattering molecular dyes to achieve specific absorption coefficients. The software facilitates recalculations of optical properties based on the actual quantities used during phantom preparation, offering flexibility and increased accuracy. Error estimates are provided using Monte Carlo sampling and error propagation. The open-source software is established as a comprehensive tool for preparing liquid turbid phantoms using microsphere suspensions, accessible to non-experts with basic familiarity of pipetting and use of analytical scales.
Structurally anisotropic materials are ubiquitous in several application fields, yet their accurate optical characterization remains challenging due our incomplete understanding of how anisotropic light transport properties arise from the microscopic scattering coefficients. In fact, even when the dynamics of light transport is directly measured, coarse simplifications are often introduced due to a lack of established theoretical models or numerical methods. Here, we apply a general Monte Carlo implementation capable of handling direction-dependent scattering to the analysis of light transport in a sample of polytetrafluoroethylene (PTFE) tape. Using only a set of transient transmittance intensity profiles, the analysis retrieves the tensor components of the diffusive rates and the scattering coefficients along all three directions, in excellent agreement with Monte Carlo simulations.
Lensless on-chip microscopy comprises a simple and compact setup in which the sample is placed close to the imaging sensor and illuminated by a coherent light source. The acquired in-line hologram carries information about the amplitude and phase image of the sample, which can be numerically reconstructed. Contrary to conventional microscopy, the reconstructed images can be numerically refocused at desired focus planes effectively providing three-dimensional information. For reliable object reconstruction, a proper focus plane must be selected, which can be done automatically using an autofocusing algorithm. The autofocusing algorithms are commonly evaluated on synthetic or experimentally acquired in-line holograms. First are usually simulated with the same numerical propagation method as used for reconstruction and are not able to simulate holograms of truly three-dimensional objects, while experimentally acquired holograms can be affected by measurement noise and model mismatch artefacts. In this paper, we propose an objective evaluation of autofocusing algorithms on in-line holograms simulated by Mie theory and T-matrix method which can simulate holograms of truly three-dimensional spherical objects distributed in various spatial positions. We evaluate and compare different autofocusing algorithms in terms of the accuracy of the estimated focus plane and computational efficiency. Finally, we present a proof-of-concept real-time implementation of the autofocusing algorithm based on the open-source PyOpenCL framework. We found that the implemented autofocusing algorithms provided the best average accuracy of 1.75 μm and required 330 μs per evaluation cycle resulting in around 20 frames per second for autofocusing a 1024×1024 hologram.
KEYWORDS: Monte Carlo methods, Computer simulations, Reflectivity, Sensors, Tissues, Transmittance, Signal to noise ratio, Optical properties, Refractive index, Multilayers
Significance: Current open-source Monte Carlo (MC) method implementations for light propagation modeling are many times tedious to build and require third-party licensed software that can often discourage prospective researchers in the biomedical optics community from fully utilizing the light propagation tools. Furthermore, the same drawback also limits rigorous cross-validation of physical quantities estimated by various MC codes.
Aim: Proposal of an open-source tool for light propagation modeling and an easily accessible dataset to encourage fruitful communications amongst researchers and pave the way to a more consistent comparison between the available implementations of the MC method.
Approach: The PyXOpto implementation of the MC method for multilayered and voxelated tissues based on the Python programming language and PyOpenCL extension enables massively parallel computation on numerous OpenCL-enabled devices. The proposed implementation is used to compute a large dataset of reflectance, transmittance, energy deposition, and sampling volume for various source, detector, and tissue configurations.
Results: The proposed PyXOpto agrees well with the original MC implementation. However, further validation reveals a noticeable bias introduced by the random number generator used in the original MC implementation.
Conclusions: Establishing a common dataset is highly important for the validation of existing and development of MC codes for light propagation in turbid media.
KEYWORDS: Monte Carlo methods, Data modeling, Reflectivity, Optical properties, Computer simulations, Spatial frequencies, Signal to noise ratio, Sensors, Scattering, Imaging systems
We propose neural network-based regression model for efficient Monte Carlo simulations of subdiffusive reflectance for spatial frequency domain imaging with low NA and validate the methodology with inverse models for estimation of optical properties.
In this work, we use a statistical skin model to compare the measured distribution of optical path lengths to time-resolved Monte Carlo simulated reflectance and validate the simulations by the use of optical phantoms.
We present a multi-layered and voxel-based Monte Carlo methods with auxiliary utilities implemented in Python for user-friendly, open-source and multi-purpose modeling of light propagation in turbid media based on PyOpenCL computational platform.
We present a simple approach to determine the refractive index of polystyrene microspheres which are frequently utilized as scatterers in turbid phantoms. The approach is based on Mie theory and transmittance measurements of polystyrene microspheres suspended in media with different refractive indices allowing simultaneous optimization of the diameter and refractive index of the polystyrene microspheres. The refractive index of the medium is changed through the addition of sucrose. Based on our preliminary results, the estimated refractive index of polystyrene microspheres deviates from the literature values by 0.2% and the estimated diameter by 20 nm from the nominal value provided by the manufacturer.
Reflectance spectroscopy shows itself as a useful tool to characterize turbid media, such as biological tissues. The light backscattered from the medium is usually collected by imaging systems or optical fiber probes. In this work we used an optical fiber probe, with a linear arrangement of the source and detection fibers that allows spatially resolved reflectance (SRR) measurements. Through the use of inverse model, the collected SRR can be exploited to estimate the optical properties of the turbid medium. The estimation process involves matching of the measured and simulated SRR that accounts for all the details of the measurement setting. At small source-detector separations and/or non-negligible absorbance, the reflectance becomes highly dependent on the scattering phase function of the medium, which can be efficiently described by the higher order Legendre moments and related scattering phase function quantifiers (PFQ). In our previous studies, we utilized the Gegenbauer Kernel (GK) scattering phase function to describe the light propagation in turbid samples. However, the domain of GK-based PFQs is quite small and fails to fully encompass the scattering phase functions of microspherical suspensions, typically used for calibration and validation of SRR measurement settings. This limitation could be overcome by utilizing scattering phase function models with a large PFQ domain that may also lead to more accurate and robust inverse model predictions. To verify this hypothesis, we evaluate various scattering phase function models that maximize the PFQ domain and experimentally validate the inverse models by SRR collected from optical phantoms and various turbid samples.
KEYWORDS: Monte Carlo methods, Reflectivity, Computer simulations, Optical properties, Scattering, Signal to noise ratio, Photon transport, Error analysis, Optical fibers, Data modeling
Monte Carlo (MC) method is regarded as the gold standard for modeling the light transport in biological tissues. Due to the stochastic nature of the MC method, many photon packets need to be processed to obtain an adequate quality of the simulated reflectance. The number of required photon packets further increases if the numerical aperture of the detection scheme is low. Consequently, extensively long simulation times may be required to obtain adequate quality of the reflectance for such detection schemes. In this paper we propose an efficient regression model that maps reflectance simulated at the maximum acceptance angle of 90◦ to the reflectance corresponding to a much smaller realistic acceptance angle. The results of validation on spatially resolved reflectance and inverse models for estimation of optical properties show that the regression models are accurate and do not introduce additional errors into the spatially resolved reflectance or the optical properties estimated by appropriate inverse models from the regressed reflectance.
In this paper, we propose a novel calibration procedure based on modeling and measurement of the reflectance distance profiles from a metallic mirror. We observe a remarkable agreement of our reflectance distance profile model with the measurements yielding repeatable calibration factors within 2% when tested on silver and aluminum mirrors. Comparison to widely acceptable calibration using polystyrene microspheres suspensions yields errors of below 10%.
Experimental setup geometry in Monte Carlo (MC) simulations is often simplified to shorten computation times. We investigate the effect of these simplifications on the accuracy of the spatial frequency domain (SFD) reflectance. We also introduce a new detection scheme in the MC method that eliminates the often overlooked errors arising from the Hankel transform of the spatially discretized reflectance profiles to SFD reflectance. Finally, we propose and evaluate an artificial neural network-based framework capable of estimating high-definition maps of optical properties in real-time.
To overcome the drawbacks of the commonly used lookup table inverse models, we propose a novel custom OpenCL™- accelerated artificial neural network inverse model for spatial frequency domain imaging (https://bitbucket.org/xopto /rftroop). Utilizing a mid-range graphics processing unit, the proposed inverse model can estimate high-definition (1920 × 1080) maps of the absorption and reduced scattering coefficients and two scattering phase function related quantifiers at a rate of more than 50 frames per second. We show that the artificial neural network inverse model can be seamlessly extended to estimate multiple optical properties independently, thus providing a versatile framework that allows introduction of new quantifiers.
Stochastic Monte Carlo method (MC) is often used to model light propagation in biological tissues. Since many photon packets need to be process to attain good quality of the simulated data, the experimental geometry in MC simulations is usually substantially simplified to shorten the computation times. However, such simplifications have been shown to introduce large simulation errors when using optical fiber probes. In our previous study, we have shown that the frequently used laterally uniform probe-sample interface simplification can introduce significant errors into the MC simulations of spatially resolved reflectance (SRR) potentially exceeding 200 %. Unfortunately, using full details of the probe tip in the MC simulations breaks down the radial symmetry of the detection scheme. Consequently, the simulation time required to obtain a good quality SRR increases by about two orders of magnitude. In this study, we introduce a framework for efficient and accurate MC simulations of SRR acquired by optical fiber probes that accounts for all the details of the probe tip including reflectance from the stainless steel and the refractive indices of the epoxy fill and optical fibers. For this purpose, we introduce an efficient regression model that maps SRR obtained through fast MC simulations based on simplified geometry to the SRR simulated by full details of the probe-sample interface. We show that a small number of SRR samples is sufficient to determine the parameters of the regression model. Finally, we use the regression model to simulate SRR for a stainless steel optical probe with six linearly placed fibers and build inverse models for estimation of absorption and reduced scattering coefficients and subdiffusive scattering phase function related quantifiers.
A measurement system was developed to acquire and analyze subdiffusive spatially resolved reflectance using an optical fiber probe with short source-detector separations. Since subdiffusive reflectance significantly depends on the scattering phase function, the analysis of the acquired reflectance is based on a novel inverse Monte Carlo model that allows estimation of phase function related parameters in addition to the absorption and reduced scattering coefficients. In conjunction with our measurement system, the model allowed real-time estimation of optical properties, which we demonstrate for a case of dynamically induced changes in human skin by applying pressure with an optical fiber probe.
Timely estimation of optical properties from spatially resolved reflectance is a challenging task since the inverse light propagation model needs to be evaluated in real time. In this paper, we propose and extensively evaluate artificial neural network based regression model for estimation of optical and structural sample properties from spatially resolved reflectance acquired by optical fiber probes. We show that the proposed regression model can be prepared from datasets of Monte Carlo simulated spatially resolved reflectance and evaluated significantly faster than the frequently used dense lookup table inverse model. We observed computation time improvements exceeding 4 orders of magnitude. Moreover, the regression model can be easily extended to estimate more free parameters without reducing the estimation accuracy. Finally, we utilized the proposed regression model to estimate optical properties of human skin subjected to dynamically changing contact pressure applied by an optical fiber probe.
We significantly improve the estimation of the absorption and reduced scattering coefficients, and second-order similarity parameter γ from spatially resolved reflectance by extending the inverse model with the third-order similarity parameter δ.
For a given experimental setting, the measured spatially resolved reflectance rapidly drops with decreasing numerical aperture of the detection scheme. Consequently, for detection schemes with small numerical apertures, the computational time of MC simulations required to obtain adequate signal-to-noise ratio of the spatially resolved reflectance can become very long. We mitigate the issue by virtually increasing the numerical aperture of the detection scheme in MC simulations and devise a criterion for robust estimation of its maximum value. By using the proposed methodology, we show that the acceptance angle of a selected imaging system can be virtually increased from 3 to 11 while preserving a low relative error of the simulated spatially resolved reflectance over a wide range of tissue-like optical properties. As a result, a more than eightfold improvement in the computation time is attained.
We propose and objectively evaluate an inverse Monte Carlo model for estimation of absorption and reduced scattering coefficients and similarity parameter γ from spatially resolved reflectance (SRR) profiles in the subdiffusive regime. The similarity parameter γ carries additional information on the phase function that governs the angular properties of scattering in turbid media. The SRR profiles at five source-detector separations were acquired with an optical fiber probe. The inverse Monte Carlo model was based on a cost function that enabled robust estimation of optical properties from a few SRR measurements without a priori knowledge about spectral dependencies of the optical properties. Validation of the inverse Monte Carlo model was performed on synthetic datasets and measured SRR profiles of turbid phantoms comprising molecular dye and polystyrene microspheres. We observed that the additional similarity parameter γ substantially reduced the reflectance variability arising from the phase function properties and significantly improved the accuracy of the inverse Monte Carlo model. However, the observed improvement of the extended inverse Monte Carlo model was limited to reduced scattering coefficients exceeding ∼15 cm−1, where the relative root-mean-square errors of the estimated optical properties were well within 10%.
Push-broom hyperspectral imaging systems are increasingly used for various medical, agricultural and military purposes. The acquired images contain spectral information in every pixel of the imaged scene collecting additional information about the imaged scene compared to the classical RGB color imaging. Due to the misalignment and imperfections in the optical components comprising the push-broom hyperspectral imaging system, variable spectral and spatial misalignments and blur are present in the acquired images. To capture these distortions, a spatially and spectrally variant response function must be identified at each spatial and spectral position. In this study, we propose a procedure to characterize the variant response function of Short-Wavelength Infrared (SWIR) push-broom hyperspectral imaging systems in the across-track and along-track direction and remove its effect from the acquired images. A custom laser-machined spatial calibration targets are used for the characterization. The spatial and spectral variability of the response function in the across-track and along-track direction is modeled by a parametrized basis function. Finally, the characterization results are used to restore the distorted hyperspectral images in the across-track and along-track direction by a Richardson-Lucy deconvolution-based algorithm. The proposed calibration method in the across-track and along-track direction is thoroughly evaluated on images of targets with well-defined geometric properties. The results suggest that the proposed procedure is well suited for fast and accurate spatial calibration of push-broom hyperspectral imaging systems.
In this paper, diffuse reflectance hyperspectral images of a light beam propagating through a semi-infinite homogeneous layer were simulated by a modified version of the open source Monte Carlo (MC) for multi-layered tissues. Subsequently, the optical properties in terms of absorption and reduced scattering coefficients were extracted from the simulated hyperspectral images by an inverse MC model based on a criterion function that exploits the spatially resolved information of hyperspectral images. The method was validated on real hyperspectral images of turbid phantoms with exactly defined optical properties.
In this paper, the commonly used inverse Monte Carlo model based on absorption and reduced scattering coefficients is extended by a well-known similarity parameter γ (gamma), which carries additional information on the phase function. Sub-diffuse reflectance measurements at five source-detector separations were used to extract the absorption and reduced scattering coefficients and phase function information encapsulated in γ. The performance of the extended inverse Monte Carlo model was evaluated by simulated and experimental reflectance spectra of turbid phantoms. A three-fold increase in the accuracy of the extended inverse Monte Carlo model that incorporates γ was observed.
We assess the properties of contact pressure applied by manually operated fiber-optic probes as a function of the operator, probe contact area, and sample stiffness. First, the mechanical properties of human skin sites with different skin structures, thicknesses, and underlying tissues were studied by in vivo indentation tests. According to the obtained results, three different homogeneous silicone skin phantoms were created to encompass the observed range of mechanical properties. The silicon phantoms were subsequently used to characterize the properties of the contact pressure by 10 experienced probe operators employing fiber-optic probes with different contact areas. A custom measurement system was used to collect the time-lapse of diffuse reflectance and applied contact pressure. The measurements were characterized by a set of features describing the transient and steady-state properties of the contact pressure and diffuse reflectance in terms of rise time, optical coupling, average value, and variability. The average applied contact pressure and contact pressure variability were found to significantly depend on the probe operator, probe contact area, and surprisingly also on the sample stiffness. Based on the presented results, we propose a set of practical guidelines for operators of manual probes.
Diffuse reflectance spectroscopy utilizing optical fiber probes is a useful and simple method for non-invasive determination of biological tissue optical properties. In order to extract the optical properties from the acquired diffuse reflectance spectra, an accurate light propagation model, such as Monte Carlo, is required. The results obtained by the model can significantly depend on the description of the tissue and optical fiber probe geometry. Optical fiber probes commonly comprise fibers arranged into a desired source-detector layout enclosed in a stainless steel ferrule. By using Monte Carlo simulations, we investigate the impact of the stainless steel optical fiber probe-tissue interface on the diffuse reflectance spectra. For this purpose, a commonly used simple laterally uniform optical probe-tissue interface with mismatched refractive indices was compared to the improved optical probe-tissue interface taking into account the fiber layout and the specular reflections from the stainless steel probe tip. The results show that the error introduced into the simulated diffuse reflectance by the simplified probe-tissue interface can easily exceed 5%.
Hyperspectral imaging systems effectively collect information across the spectral and two spatial dimensions by employing three main components: the front lens, the light-diffraction element and a camera. Imperfections in these components introduce spectral and spatial dependent distortions in the recorded hyperspectral image. These can be characterized by a 3D response function that is subsequently used to remove distortions and enhance the resolution of the recorded images by deconvolution. The majority of existing characterization methods assume spatial and spectral separability of the 3D response function. In this way, the complex problem of 3D response function characterization is reduced to independent characterizations of the three orthogonal response function components. However, if the 3D response function is non-separable, such characterization can lead to poor response function estimates, and hence inaccurate and distorted results of the subsequent deconvolution-based calibration and image enhancement. In this paper, we evaluate the influence of the spatial response function non-separability on the results of the calibration by deconvolution. For this purpose, a novel procedure for direct measurement of the 2D spatial response function is proposed along with a quantitative measure of the spatial response function non-separability. The quality of deconvolved images is assessed in terms of full width at half maximum (FWHM) and step edge overshoot magnitude observed in the deconvolved images of slanted edges, images of biological slides, and 1951 USAF resolution test chart. Results show that there are cases, when nonseparability of the system response function is significant and should be considered by the deconvolution-based calibration and image enhancement methods.
In this study, we assess the properties of the contact pressure applied by manually operated probes as a function of the operator, probe contact area, and sample stiffness. For this purpose, three different human skin-like phantoms with different well-defined mechanical properties were used. To gain relevant statistics of the contact pressure properties, the study included ten experienced probe operators that were asked to apply gentle contact pressure to the skin-like phantoms by using three different contact area probes. A novel system for rapid simultaneous acquisition of spectra and corresponding contact pressure was used to collect the relevant information. Results show that the variability of the gentle contact pressure significantly depends on the probe operator, sample stiffness and probe contact area.
Diffuse reflectance spectroscopy is a popular approach for non-invasive assessment of optical properties in turbid media. The acquired spectra are analyzed by various light propagation models or by purely empirical methods. In this study, we quantitatively asses the experimental data and Monte Carlo lookup table-based inverse models by extracting the optical properties from the diffuse reflectance spectra of two carefully prepared turbid phantom sets with exactly defined optical properties. The first turbid phantom set was used for the creation of the experimental data-based lookup table model and calibration of the Monte Carlo lookup table-based inverse model. The second phantom set was used for the evaluation and comparative assessment of the two lookup table-based inverse models. In addition, we investigate the possible sources of errors introduced by the inverse models and show that the lookup table-based models disregard important information regarding the medium scattering phase function.
Cancer is the main cause of canine morbidity and mortality. The existing evaluation of tumors requires an experienced veterinarian and usually includes invasive procedures (e.g., fine-needle aspiration) that can be unpleasant for the dog and the owner. We investigate visible and near-infrared diffuse reflectance spectroscopy (DRS) as a noninvasive optical technique for evaluation and detection of canine skin and subcutaneous tumors ex vivo and in vivo. The optical properties of tumors and skin were calculated in a spectrally constrained manner, using a lookup table-based inverse model. The obtained optical properties were analyzed and compared among different tumor groups. The calculated parameters of the absorption and reduced scattering coefficients were subsequently used for detection of malignant skin and subcutaneous tumors. The detection sensitivity and specificity of malignant tumors ex vivo were 90.0% and 73.5%, respectively, while corresponding detection sensitivity and specificity of malignant tumors in vivo were 88.4% and 54.6%, respectively. The obtained results show that the DRS is a promising noninvasive optical technique for detection and classification of malignant and benign canine skin and subcutaneous tumors. The method should be further investigated on tumors with common origin.
Review of the existing studies on the contact pressure–induced changes in the optical properties of biological tissues showed that the reported changes in transmittance, reflectance, absorption, and scattering coefficient are vastly inconsistent. In order to gain more insight into the contact pressure–induced changes observed in biomedical applications involving common probe-spectrometer diffuse reflectance measurement setups and provide a set of practical guidelines minimizing the influence of the changes on the analysis of acquired spectra, we conducted a series of in vivo measurements, where the contact pressure was precisely controlled, and the spectral and contact pressure information were acquired simultaneously. Classification of three measurement sites on a human hand, representing the natural variability in the perfusion and structure of the underlying tissue, was assessed by training and evaluating classifiers at different contact pressure levels and for different probe operators. Based on the results, three practical guidelines have been proposed to avoid classification performance degradation. First, the most suitable pressure level should be identified. Second, the pressure level should be kept in a narrow range during the acquisition of spectra. Third, applications utilizing probes equipped with a calibrated spring can use several classifiers trained at different contact pressure levels to improve classification performance.
Push-broom hyperspectral imaging system ideally disperses the spectral and spatial information in two orthogonal directions preferably aligned with the columns and rows of the imaging sensor. Due to the imperfections in the camera lens and in particular the optical components of spectrograph, wavelength dependent spectral and spatial distortions along with spatial and spectral blur are introduced in the recorded image. In this study, we propose and evaluate a novel method for characterization and resolution enhancement of push-broom hyperspectral imaging systems. First, the spatially and spectrally dependent response function is characterized by measuring the response of the system to spectral and spatial reference objects. The relevant variability of the response function in the imaging plane is captured by a global parametric model. Finally, the response function estimate is used to remove distortions and enhance the spectral and spatial resolution of the system. The resolution enhancements were assessed by observing the change in full width at half-maximum of spectral response function and rise width of the spatial response function. The results of validation show that the proposed method affectively removes geometric distortions and significantly enhances the spectral and spatial resolution of the recorded images.
KEYWORDS: In vivo imaging, Diffuse reflectance spectroscopy, Skin, Spectroscopy, Tissue optics, Tissues, Control systems, Near infrared, Optical properties, Calibration
Contact pressure induced by manually operated fiber optic probes can significantly affect the optical properties of the studied tissue. If the contact pressure and the changes in optical properties are measured properly, then the complementary information can be used to obtain additional insight into the tissue physiology. However, as reliable assessment of the contact pressure in the existing diffuse reflectance setups is difficult, the impact of contact pressure is usually neglected or considered as a source of errors. We introduce a measurement system for controlled application of contact pressure and for the acquisition of diffuse reflectance spectra, which is suitable for in vivo studies and for overcoming the limitations of the existing measurement setups. A spectral-contact-pressure plane is proposed to present the combined information, highlighting the unique tissue response to the applied pressure.
Acquiring near infrared spectra in vivo usually requires a fiber-optic probe to be pressed against the tissue. The applied pressure can significantly affect the optical properties of the underlying tissue, and thereby the acquired spectra. The existing studies consider these effects to be distortions. In contrast, we hypothesize that the pressure-induced spectral response is site- and tissue-specific, providing additional information for the tissue classification. For the purpose of this study, a custom system was designed for dynamic pressure control and rapid acquisition of spectra. The pressure-induced spectral response was studied at three proximate skin sites of the human hand. The diffuse reflectance and scattering were found to decrease with the applied contact pressure. In contrast, the concentrations of chromophores, and consequently the absorption, increased with the applied contact pressure. The pressure-induced changes in the tissue optical properties were found to be site-specific and were modeled as a polynomial function of the applied contact pressure. A quadratic discriminant analysis classification of the tissue spectra acquired at the three proximate skin sites, based on the proposed pressure-induced spectral response model, resulted in a high (90%) average classification sensitivity and specificity, clearly supporting the working hypothesis.
Spatial resolution of hyperspectral imaging systems can vary significantly due to axial optical aberrations that
originate from wavelength-induced index-of-refraction variations of the imaging optics. For systems that have a
broad spectral range, the spatial resolution will vary significantly both with respect to the acquisition wavelength
and with respect to the spatial position within each spectral image. Variations of the spatial resolution can be
effectively characterized as part of the calibration procedure by a local image-based estimation of the pointspread
function (PSF) of the hyperspectral imaging system. The estimated PSF can then be used in the image
deconvolution methods to improve the spatial resolution of the spectral images. We estimated the PSFs from
the spectral images of a line grid geometric caliber. From individual line segments of the line grid, the PSF was
obtained by a non-parametric estimation procedure that used an orthogonal series representation of the PSF.
By using the non-parametric estimation procedure, the PSFs were estimated at different spatial positions and
at different wavelengths. The variations of the spatial resolution were characterized by the radius and the fullwidth
half-maximum of each PSF and by the modulation transfer function, computed from images of USAF1951
resolution target. The estimation and characterization of the PSFs and the image deconvolution based spatial
resolution enhancement were tested on images obtained by a hyperspectral imaging system with an acousto-optic
tunable filter in the visible spectral range. The results demonstrate that the spatial resolution of the acquired
spectral images can be significantly improved using the estimated PSFs and image deconvolution methods.
Hyper-spectral imaging has gained recognition as an important non-invasive research tool in the field of biomedicine.
Among the variety of available hyperspectral imaging systems, systems comprising an imaging spectrograph, lens, wideband
illumination source and a corresponding camera stand out for the short acquisition time and good signal to noise
ratio. The individual images acquired by imaging spectrograph-based systems contain full spectral information along one
spatial dimension. Due to the imperfections in the camera lens and in particular the optical components of the imaging
spectrograph, the acquired images are subjected to spatial and spectral distortions, resulting in scene dependent nonlinear
spectral degradations and spatial misalignments which need to be corrected. However, the existing correction methods
require complex calibration setups and a tedious manual involvement, therefore, the correction of the distortions is often
neglected. Such simplified approach can lead to significant errors in the analysis of the acquired hyperspectral images. In
this paper, we present a novel fully automated method for correction of the geometric and spectral distortions in the
acquired images. The method is based on automated non-rigid registration of the reference and acquired images
corresponding to the proposed calibration object incorporating standardized spatial and spectral information. The
obtained transformation was successfully used for sub-pixel correction of various hyperspectral images, resulting in
significant improvement of the spectral and spatial alignment. It was found that the proposed calibration is highly
accurate and suitable for routine use in applications involving either diffuse reflectance or transmittance measurement
setups.
Despite major improvements in dental healthcare and technology, dental caries remains one of the most prevalent
chronic diseases of modern society. The initial stages of dental caries are characterized by demineralization of enamel
crystals, commonly known as white spots, which are difficult to diagnose. Near-infrared (NIR) hyperspectral imaging is
a new promising technique for early detection of demineralization which can classify healthy and pathological dental
tissues. However, due to non-ideal illumination of the tooth surface the hyperspectral images can exhibit specular
reflections, in particular around the edges and the ridges of the teeth. These reflections significantly affect the
performance of automated classification and visualization methods. Cross polarized imaging setup can effectively
remove the specular reflections, however is due to the complexity and other imaging setup limitations not always
possible. In this paper, we propose an alternative approach based on modeling the specular reflections of hard dental
tissues, which significantly improves the classification accuracy in the presence of specular reflections. The method was
evaluated on five extracted human teeth with corresponding gold standard for 6 different healthy and pathological hard
dental tissues including enamel, dentin, calculus, dentin caries, enamel caries and demineralized regions. Principal
component analysis (PCA) was used for multivariate local modeling of healthy and pathological dental tissues. The
classification was performed by employing multiple discriminant analysis. Based on the obtained results we believe the
proposed method can be considered as an effective alternative to the complex cross polarized imaging setups.
Accurate and objective evaluation of vertebral body deformations represents an important part of the clinical diagnostics
and therapy of pathological conditions affecting the spine. Although modern clinical practice is oriented towards threedimensional
(3D) imaging techniques, the established methods for the evaluation of vertebral body deformations are
based on measurements in two-dimensional (2D) X-ray images. In this paper, we propose a method for automatic
measurement of vertebral body deformations in computed tomography (CT) images that is based on efficient modeling
of the vertebral body shape with a 3D parametric model. By fitting the 3D model to the vertebral body in the image,
quantitative description of normal and pathological vertebral bodies is obtained from the value of 25 parameters of the
model. The evaluation of vertebral body deformations is based on the distance of the observed vertebral body from the
distribution of the parameter values of normal vertebral bodies in the parametric space. The distribution is obtained from
80 normal vertebral bodies in the training data set and verified with eight normal vertebral bodies in the control data set.
The statistically meaningful distance of eight pathological vertebral bodies in the study data set from the distribution of
normal vertebral bodies in the parametric space shows that the parameters can be used to successfully model vertebral
body deformations in 3D. The proposed method may therefore be used to assess vertebral body deformations in 3D or
provide clinically meaningful observations that are not available when using 2D methods that are established in clinical
practice.
KEYWORDS: Tissues, Skin, Near infrared, In vitro testing, In vivo imaging, Reflectivity, Tissue optics, Diffuse reflectance spectroscopy, Spectroscopy, Principal component analysis
Near infrared (NIR) diffuse reflectance spectroscopy is a rapid noninvasive spectroscopic technique extensively used in
the field of biomedicine. The main challenges of the NIR spectroscopy lie in the analysis of the acquired spectral
information, requiring prior knowledge on the influence of various tissue parameters on the spectra. In this paper, we
focus on the influence of pressure on the soft tissues NIR spectra. The existing studies based on static measurements
showed that pressure tends to decrease the diffuse reflectance of the soft tissue. The purpose of this study was to further
analyze the effects of static and in particular dynamically changing pressure on the soft tissue diffuse reflectance
properties and assess its potential for tissue classification. For this purpose, a custom real-time system for rapid
acquisition of NIR diffuse reflectance spectra of tissue under controlled static or dynamically changing pressure was
developed. The study was conducted using porcine ribs and porcine liver (in vitro) tissue samples and human skin (in
vivo). The acquired spectra were preprocessed by mean centering or standard normal variate and analyzed by Principal
Component Analysis (PCA). A polynomial function was used to model the first three calculated scores as a function of
the applied pressure. It was found, that the median values of the calculated polynomial coefficients were significantly
different for all the samples, indicating reflectance changes due to the applied pressure could be used for classification of
biological tissues.
Despite major improvements in dental healthcare and oral hygiene, dental caries remains one of the most prevalent oral
diseases and represents the primary cause of oral pain and tooth loss. The initial stages of dental caries are characterized
by demineralization of enamel crystals and are difficult to diagnose. Near infrared (NIR) hyperspectral imaging is a new
promising technique for detection of early changes in the surfaces of carious teeth. This noninvasive imaging technique
can characterize and differentiate between the sound tooth surface and initial or advanced tooth caries. The absorbing
and scattering properties of dental tissues reflect in distinct spectral features, which can be measured, quantified and used
to accurately classify and map different dental tissues. Specular reflections from the tooth surface, which appear as bright
spots, mostly located around the edges and the crests of the teeth, act as a noise factor which can significantly interfere
with the spectral measurements and analysis of the acquired images, degrading the accuracy of the classification and
diagnosis. Employing cross-polarized imaging setup can solve this problem, however has yet to be systematically
evaluated, especially in broadband hyperspectral imaging setups. In this paper, we employ cross-polarized illumination
setup utilizing state-of-the-art high-contrast broadband wire-grid polarizers in the spectral range from 900 nm to 1700
nm for hyperspectral imaging of natural and artificial carious lesions of various degrees.
Dental caries is a disease characterized by demineralization of enamel crystals leading to the penetration of bacteria into
the dentine and pulp. Early detection of enamel demineralization resulting in increased enamel porosity, commonly
known as white spots, is a difficult diagnostic task. Laser induced autofluorescence was shown to be a useful method for
early detection of demineralization. The existing studies involved either a single point spectroscopic measurements or
imaging at a single spectral band. In the case of spectroscopic measurements, very little or no spatial information is
acquired and the measured autofluorescence signal strongly depends on the position and orientation of the probe. On the
other hand, single-band spectral imaging can be substantially affected by local spectral artefacts. Such effects can
significantly interfere with automated methods for detection of early caries lesions. In contrast, hyperspectral imaging
effectively combines the spatial information of imaging methods with the spectral information of spectroscopic methods
providing excellent basis for development of robust and reliable algorithms for automated classification and analysis of
hard dental tissues. In this paper, we employ 405 nm laser excitation of natural caries lesions. The fluorescence signal is
acquired by a state-of-the-art hyperspectral imaging system consisting of a high-resolution acousto-optic tunable filter
(AOTF) and a highly sensitive Scientific CMOS camera in the spectral range from 550 nm to 800 nm. The results are
compared to the contrast obtained by near-infrared hyperspectral imaging technique employed in the existing studies on
early detection of dental caries.
Construction of a standardized near infrared (NIR) hyper-spectral teeth database is a first step in the development
of a reliable diagnostic tool for quantification and early detection of dental diseases. The standardized diffuse
reflectance hyper-spectral database was constructed by imaging 12 extracted human teeth with natural lesions of
various degrees in the spectral range from 900 to 1700 nm with spectral resolution of 10 nm. Additionally, all the
teeth were imaged by X-ray and digital color camera. The color and X-ray teeth images were presented to the
expert for localization and classification of the dental diseases, thereby obtaining a dental disease gold standard.
Accurate transfer of the dental disease gold standard to the NIR images was achieved by image registration in a
groupwise manner, taking advantage of the multichannel image information and promoting image edges as the
features for the improvement of spatial correspondence detection. By the presented fully automatic multi-modal
groupwise registration method, images of new teeth samples can be accurately and reliably registered and then
added to the standardized NIR hyper-spectral teeth database. Adding more samples increases the biological
and patho-physiological variability of the NIR hyper-spectral teeth database and can importantly contribute to
the objective assessment of the sensitivity and specificity of multivariate image analysis techniques used for the
detection of dental diseases. Such assessment is essential for the development and validation of reliable qualitative
and especially quantitative diagnostic tools based on NIR spectroscopy.
Despite major improvements in dental healthcare and technology, dental caries remains one of the most prevalent
chronic diseases of modern society. The initial stages of dental caries are characterized by demineralization of enamel
crystals, commonly known as white spots which are difficult to diagnose. If detected early enough, such
demineralization can be arrested and reversed by non-surgical means through well established dental treatments (fluoride
therapy, anti-bacterial therapy, low intensity laser irradiation). Near-infrared (NIR) hyper-spectral imaging is a new
promising technique for early detection of demineralization based on distinct spectral features of healthy and
pathological dental tissues. In this study, we apply NIR hyper-spectral imaging to classify and visualize healthy and
pathological dental tissues including enamel, dentin, calculus, dentin caries, enamel caries and demineralized areas. For
this purpose, a standardized teeth database was constructed consisting of 12 extracted human teeth with different degrees
of natural dental lesions imaged by NIR hyper-spectral system, X-ray and digital color camera. The color and X-ray
images of teeth were presented to a clinical expert for localization and classification of the dental tissues, thereby
obtaining the gold standard. Principal component analysis was used for multivariate local modeling of healthy and
pathological dental tissues. Finally, the dental tissues were classified by employing multiple discriminant analysis. High
agreement was observed between the resulting classification and the gold standard with the classification sensitivity and
specificity exceeding 85 % and 97 %, respectively. This study demonstrates that NIR hyper-spectral imaging has
considerable diagnostic potential for imaging hard dental tissues.
Dental caries is a disease characterized by demineralization of enamel crystals leading to the penetration of bacteria into
the dentin and pulp. If left untreated, the disease can lead to pain, infection and tooth loss. Early detection of enamel
demineralization resulting in increased enamel porosity, commonly known as white spots, is a difficult diagnostic task.
Several papers reported on near infrared (NIR) spectroscopy to be a potentially useful noninvasive spectroscopic
technique for early detection of caries lesions. However, the conducted studies were mostly qualitative and did not
include the critical assessment of the spectral variability of the sound and carious dental tissues and influence of the
water content. Such assessment is essential for development and validation of reliable qualitative and especially
quantitative diagnostic tools based on NIR spectroscopy. In order to characterize the described spectral variability, a
standardized diffuse reflectance hyper-spectral database was constructed by imaging 12 extracted human teeth with
natural lesions of various degrees in the spectral range from 900 to 1700 nm with spectral resolution of 10 nm.
Additionally, all the teeth were imaged by digital color camera. The influence of water content on the acquired spectra
was characterized by monitoring the teeth during the drying process. The images were assessed by an expert, thereby
obtaining the gold standard. By analyzing the acquired spectra we were able to accurately model the spectral variability
of the sound dental tissues and identify the advantages and limitations of NIR hyper-spectral imaging.
Optical aberrations present an important problem in optical measurements. Geometrical calibration of an imaging system
is therefore of the utmost importance for achieving accurate optical measurements. In hyper-spectral imaging systems,
the problem of optical aberrations is even more pronounced because optical aberrations are wavelength dependent. Geometrical
calibration must therefore be performed over the entire spectral range of the hyper-spectral imaging system,
which is usually far greater than that of the visible light spectrum. This problem is especially adverse in AOTF (Acousto-
Optic Tunable Filter) hyper-spectral imaging systems, as the diffraction of light in AOTF filters is dependent on both
wavelength and angle of incidence. Geometrical calibration of hyper-spectral imaging system was performed by stable
caliber of known dimensions, which was imaged at different wavelengths over the entire spectral range. The acquired
images were then automatically registered to the caliber model by both parametric and nonparametric transformation
based on B-splines and by minimizing normalized correlation coefficient. The calibration method was tested on an
AOTF hyper-spectral imaging system in the near infrared spectral range. The results indicated substantial wavelength
dependent optical aberration that is especially pronounced in the spectral range closer to the infrared part of the spectrum.
The calibration method was able to accurately characterize the aberrations and produce transformations for efficient
sub-pixel geometrical calibration over the entire spectral range, finally yielding better spatial resolution of hyperspectral
imaging system.
Near-infrared spectroscopy is a promising, rapidly developing, reliable and noninvasive technique, used extensively in
the biomedicine and in pharmaceutical industry. With the introduction of acousto-optic tunable filters (AOTF) and
highly sensitive InGaAs focal plane sensor arrays, real-time high resolution hyper-spectral imaging has become feasible
for a number of new biomedical in vivo applications. However, due to the specificity of the AOTF technology and lack
of spectral calibration standardization, maintaining long-term stability and compatibility of the acquired hyper-spectral
images across different systems is still a challenging problem. Efficiently solving both is essential as the majority of
methods for analysis of hyper-spectral images relay on a priori knowledge extracted from large spectral databases,
serving as the basis for reliable qualitative or quantitative analysis of various biological samples. In this study, we
propose and evaluate fast and reliable spectral calibration of hyper-spectral imaging systems in the short wavelength
infrared spectral region. The proposed spectral calibration method is based on light sources or materials, exhibiting
distinct spectral features, which enable robust non-rigid registration of the acquired spectra. The calibration accounts for
all of the components of a typical hyper-spectral imaging system such as AOTF, light source, lens and optical fibers. The
obtained results indicated that practical, fast and reliable spectral calibration of hyper-spectral imaging systems is
possible, thereby assuring long-term stability and inter-system compatibility of the acquired hyper-spectral images.
Visualization of subcutaneous veins is very difficult with the naked eye, but important for diagnosis of medical
conditions and different medical procedures such as catheter insertion and blood withdrawal. Moreover, recent studies
showed that the images of subcutaneous veins could be used for biometric identification. The majority of methods used
for enhancing the contrast between the subcutaneous veins and surrounding tissue are based on simple imaging systems
utilizing CMOS or CCD cameras with LED illumination capable of acquiring images from the near infrared spectral
region, usually near 900 nm. However, such simplified imaging methods cannot exploit the full potential of the spectral
information. In this paper, a new highly versatile method for enhancing the contrast of subcutaneous veins based on
state-of-the-art high-resolution hyper-spectral imaging system utilizing the spectral region from 550 to 1700 nm is
presented. First, a detailed analysis of the contrast between the subcutaneous veins and the surrounding tissue as a
function of wavelength, for several different positions on the human arm, was performed in order to extract the spectral
regions with the highest contrast. The highest contrast images were acquired at 1100 nm, however, combining the
individual images from the extracted spectral regions by the proposed contrast enhancement method resulted in a single
image with up to ten-fold better contrast. Therefore, the proposed method has proved to be a useful tool for visualization
of subcutaneous veins.
The unique properties of light emitting diodes (LED) offer significant advantages in terms of lifetime, source control,
and response time all of which are potentially very important for various illumination applications. However, a number
of factors such as the initial variability of the optical properties in a batch, high temperature dependence of the intensity
and color especially of the red and amber LEDs and spectral degradation causing gradual intensity decrease and shift in
color, need to be considered when developing LED light sources. The degradation and temperature dependence of the
LEDs can significantly affect the performance of a light source over the operating period in terms of intensity and color
stability, and efficacy. This is especially critical for polychromatic LED light sources, consisting of different types of
LEDs with different spectral degradation rates and temperature dependencies. Therefore, a method estimating the
intensity and color as a function of the affecting factors could be a valuable tool for developing LED light sources for
various applications. In this paper, we present a simulator employing model-based estimation of the LED color, intensity,
and efficacy as a function of temperature, forward current, and spectral degradation. The models are derived from the
measurements of the electrical and spectral characteristic of the individual LEDs as a function of the forward current,
junction temperature and corresponding thermal properties of the packaging. The results show that the simulator
provides accurate estimates of the intensity and color of the LED light sources as a function of the temperature and
forward current.
The unique properties of LEDs offer significant advantages in terms of lifetime, intensity and color control, response
time and efficiency, all of which are very important for illumination in machine vision applications. However, there are
some drawbacks of LEDs, such as the high thermal dependency and temporal degradation of the intensity and color.
Dealing with these drawbacks requires complex LED drivers, which are able to compensate for the abovementioned
changes in the intensity and color, thereby maintaining higher stability over a wide range of ambient temperature
throughout the lifetime of a LED light source. Moreover, state-of-the-art machine vision systems usually consist of a
large number of independent LED light sources that enable real-time switching between different illumination setups at
frequencies of up to 100 kHz. In this paper, we discuss the concepts of smart LED drivers with the emphasis on the
flexibility and applicability. All the most important characteristics are being considered and discussed in detail: the
accurate generation of high frequency waveforms, the efficiency of the current driver, thermal and temporal stabilization
of the LED intensity and color, communication with a camera and personal computer or embedded system, and
practicalities of implementing a large number of independent drive channels. Finally, a practical solution addressing all
of the abovementioned issues is proposed with the aim of providing a flexible and highly stable smart LED light source
driver for state-of-the-art machine vision systems.
In this paper, a current-accelerated method for fast estimation of the intensity and color degradation with corresponding
variability of light-emitting diodes (LEDs) is presented. The method is based on automated periodical spectral
measurements of emitted light of LEDs under different current loads, which include nominal and several different above
nominal currents. The intensity and color degradation and corresponding variability among different LEDs at a specified
current are estimated by detailed analysis of the acquired spectral data at above nominal currents. The method was
validated by comparing the predicted values with the measured values at nominal current. The proposed method was
tested on white LEDs (luminous intensity ranged from 5 to 20 cd) from five manufacturers (Nichia, Etgtech, Sansen,
Daina, and Velleman). The results show that the intensity and color degradation and corresponding variability among
different LEDs is significant and, therefore, should be considered with great care when designing highly demanding
lighting products.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.