Deconvolution techniques for experimental optical imaging in medicine

Optical imaging in biomedicine is essentially determined by the light absorption and scattering interaction of microscopic and macroscopic constituents in the medium. In the near infrared range (NIR) the scattering event, with a coefficient approximately two orders of magnitude greater than absorption, plays a dominant role. The size of the detected scattered light volume is apparently larger than the original one. The purpose of this paper is to quantify these effects for fluorescent light. To this effect, the optical properties of dental composite phantoms were measured and the blurring of a point source – referred to as generalized point spread function (PSF) – was calculated by Monte Carlo simulations. We demonstrate the use of this method for scaling the concentration of fluorophores and for reconstruction of the true shape of fluorescence regions below the scattering layer of the phantom by deconvolution. The combination of these two mathematical techniques – construction of the PSF by Monte Carlo simulation (MCS) and deconvolution of the stray light picture – is of benefit for the interpretation of fluorescence and scattering images used for various medical applications.     

Application of hyperspectral imaging technology for rapid identification of Ruditapes philippinarum contaminated by heavy metals

Human beings are confronted with a serious health hazard when ingesting Ruditapes philippinarum contaminated with heavy metals, and thus it is significantly necessary to identify heavy metal contaminated Ruditapes philippinarum. This study investigates the feasibility of hyperspectral imaging to identify heavy metal contamination in Ruditapes philippinarum rapidly. To reduce the effects of noise, four different spectral pretreatments were performed on the original spectra. To select characteristic wavebands for identification, four waveband selection algorithms based on neighbourhood rough set theory were proposed, namely, mutual information, consistency measure, dependency measure, and variable precision. The selected wavebands were input to an extreme learning machine to construct classification models. The results demonstrated that multiplicative scatter correction pretreatment was suitable for Ruditapes philippinarum hyperspectral imaging datasets. The identification models exhibited satisfactory performance to distinguish healthy Ruditapes philippinarum from those contaminated by both individual and multiple heavy metals. The identification results of Cd and Pb contaminated samples were more accurate than those of Cu and Zn contaminated samples. When the number of training samples decreased the identification performance decreased, but not significantly. The results showed that combined with pattern recognition analysis hyperspectral imaging technology can be used to distinguish healthy Ruditapes philippinarum samples from those contaminated by heavy metals, even with only a small number of training samples. This model is suitable for applications in analysing many shellfish rapidly and non-destructively.

Combined with pattern recognition analysis hyperspectral imaging technology can be used to identify heavy metal contamination in Ruditapes philippinarum rapidly and non-destructively, even with only a small number of training samples.     

Designing microarray phantoms for hyperspectral imaging validation

The design and fabrication of custom-tailored microarrays for use as phantoms in the characterization of hyperspectral imaging systems is described. Corresponding analysis methods for biologically relevant samples are also discussed. An image-based phantom design was used to program a microarrayer robot to print prescribed mixtures of dyes onto microscope slides. The resulting arrays were imaged by a hyperspectral imaging microscope. The shape of the spots results in significant scattering signals, which can be used to test image analysis algorithms. Separation of the scattering signals allowed elucidation of individual dye spectra. In addition, spectral fitting of the absorbance spectra of complex dye mixtures was performed in order to determine local dye concentrations. Such microarray phantoms provide a robust testing platform for comparisons of hyperspectral imaging acquisition and analysis methods.     

Application and interpretation of deep learning methods for the geographical origin identification of Radix Glycyrrhizae using hyperspectral imaging

Radix Glycyrrhizae is used as a functional food and traditional medicine. The geographical origin of Radix Glycyrrhizae is a determinant factor influencing the chemical and physical properties as well as its medicinal and health effects. The visible/near-infrared (Vis/NIR) (376–1044 nm) and near-infrared (NIR) hyperspectral imaging (915–1699 nm) were used to identify the geographical origin of Radix Glycyrrhizae. Convolutional neural network (CNN) and recurrent neural network (RNN) models in deep learning methods were built using extracted spectra, with logistic regression (LR) and support vector machine (SVM) models as comparisons. For both spectral ranges, the deep learning methods, LR and SVM all exhibited good results. The classification accuracy was over 90% for the calibration, validation, and prediction sets by the LR, CNN, and RNN models. Slight differences in classification performances existed between the two spectral ranges. Further, interpretation of the CNN model was conducted to identify the important wavelengths, and the wavelengths with high contribution rates that affected the discriminant analysis were consistent with the spectral differences. Thus, the overall results illustrate that hyperspectral imaging with deep learning methods can be used to identify the geographical origin of Radix Glycyrrhizae, which provides a new basis for related research.

Hyperspectral imaging provides an effective way to identify the geographical origin of Radix Glycyrrhizae to assess its quality.     

High-speed hyperspectral Raman imaging for label-free compositional microanalysis

We present high-speed hyperspectral Raman imaging with integrated active-illumination for label-free compositional microanalysis. We show that high-quality Raman spectra can be acquired from as many as ~1,000 spots/sec semi-randomly distributed among a ~100x100 μm² area without mechanical scanning. We demonstrate rapid data acquisition from three types of samples: 1) uniform, strong Raman scatterers, e.g., silicon substrates; 2) non-uniform, medium-strength Raman scatterers, e.g., polymer microparticles; and, 3) non-uniform, relatively weak Raman scatterers, e.g., bacterial spores. We compare the system performance to that of point-scan with an electron-multiplied CCD camera, as implemented in some commercial systems. The results suggest that our system not only provides significant imaging speed advantage for various types of samples, but also permits substantially longer integration time per spot, leading to superior signal-to-noise ratio data. Our system enables the rapid collection of high quality Raman spectra for reliable and robust compositional microanalysis that are potentially transformative in applications such as semiconductor material and device, polymer blend and biomedicine.OCIS codes: (180.5655) Raman microscopy, (170.1530) Cell analysis, (170.0110) Imaging systems     

Developing digital tissue phantoms for hyperspectral imaging of ischemic wounds

Hyperspectral imaging has the potential to achieve high spatial resolution and high functional sensitivity for non-invasive assessment of tissue oxygenation. However, clinical acceptance of hyperspectral imaging in ischemic wound assessment is hampered by its poor reproducibility, low accuracy, and misinterpreted biology. These limitations are partially caused by the lack of a traceable calibration standard. We proposed a digital tissue phantom (DTP) platform for quantitative calibration and performance evaluation of spectral wound imaging devices. The technical feasibility of such a DTP platform was demonstrated by both in vitro and in vivo experiments. The in vitro DTPs were developed based on a liquid blood phantom model. The in vivo DTPs were developed based on a porcine ischemic skin flap model. The DTPs were projected by a Hyperspectral Image Projector (HIP) with high fidelity. A wide-gap 2nd derivative oxygenation algorithm was developed to reconstruct tissue functional parameters from hyperspectral measurements. In this study, we have demonstrated not only the technical feasibility of using DTPs for quantitative calibration, evaluation, and optimization of spectral imaging devices but also its potential for ischemic wound assessment in clinical practice.     

Spectral-profile-based algorithm for hemoglobin oxygen saturation determination from diffuse reflectance spectra

Variations of hemoglobin (Hb) oxygenation in tissue provide important indications concerning the physiological conditions of tissue, and the data related to these variations are of intense interest in medical research as well as in clinical care. In this study, we derived a new algorithm to estimate Hb oxygenation from diffuse reflectance spectra. The algorithm was developed based on the unique spectral profile differences between the extinction coefficient spectra of oxy-Hb and deoxy-Hb within the visible wavelength region. Using differential wavelet transformation, these differences were quantified using the locations of certain spectral features, and, then, they were related to the oxygenation saturation level of Hb. The applicability of the algorithm was evaluated using a set of diffuse reflectance spectra produced by a Monte Carlo simulation model of photon migration and by tissue phantoms experimentally. The algorithm was further applied to the diffuse reflectance spectra acquired from in vivo experiments to demonstrate its clinical utility. The validation and evaluation results concluded that the algorithm is applicable to various tissue types (i.e., scattering properties) and can be easily used in conjunction with a diverse range of probe geometries for real-time monitoring of Hb oxygenation.     

Band selection for oxygenation estimation with multispectral/hyperspectral imaging

Multispectral imaging provides valuable information on tissue composition such as hemoglobin oxygen saturation. However, the real-time application of this technique in interventional medicine can be challenging due to the long acquisition times needed for large amounts of hyperspectral data with hundreds of bands. While this challenge can partially be addressed by choosing a discriminative subset of bands, the band selection methods proposed to date are mainly restricted by the availability of often hard to obtain reference measurements. We address this bottleneck with a new approach to band selection that leverages highly accurate Monte Carlo (MC) simulations. We hypothesize that a so chosen small subset of bands can reproduce or even improve upon the results of a quasi continuous spectral measurement. We further investigate whether novel domain adaptation techniques can address the inevitable domain shift stemming from the use of simulations. Initial results based on in silico and in vivo experiments suggest that 10-20 bands are sufficient to closely reproduce results from spectral measurements with 101 bands in the 500-700 nm range. The investigated domain adaptation technique, which only requires unlabeled in vivo measurements, yielded better results than the pure in silico band selection method. Overall, our method could guide development of fast multispectral imaging systems suited for interventional use without relying on complex hardware setups or manually labeled data.     

Shortwave-infrared (SWIR) emitting annexin V for high-contrast fluorescence molecular imaging of tumor apoptosis in living mice

Recently, shortwave infrared (SWIR) fluorescence imaging over 1000 nm has attracted much attention for in vivo optical imaging because of the higher signal to background ratios in the SWIR region. For the application of SWIR fluorescence imaging to biomedical fields, the development of SWIR fluorescent molecular probes with high biocompatibility is crucial. Although many researchers have designed a variety of SWIR emitting probes based on organic dyes, the synthesis of biocompatible SWIR fluorescent molecular imaging probes is still challenging. In this work we synthesized indocyanine green (ICG) and π-conjugation extended ICG (ICG-C11) labelled annexin V as SWIR fluorescent probes for tumor apoptosis. Annexin V is an endogenous protein with binding ability to phosphatidylserine (PS) which appears on the outer monolayer of apoptotic cell membranes. Although there are many types of visible and NIR fluorescent annexin V, there are no SWIR emitting fluorescent probes that can be used for high contrast fluorescence imaging of apoptosis in vivo. Herein, we report the synthesis and application of ICG and ICG-C11 conjugated annexin V for SWIR fluorescence imaging of tumor apoptosis. The presented fluorescent annexin V is the first SWIR emitting probe for in vivo optical imaging of tumor apoptosis. We demonstrate that SWIR emitting ICG- and ICG-C11 conjugated annexin V enable high-contrast fluorescence imaging of tumor apoptosis in living mice. We further demonstrate that ICG-C11-annexin V can be used for long-term (ca. two weeks) SWIR fluorescence imaging of tumor apoptosis. The SWIR fluorescent annexin V will greatly contribute not only to the study of tumor-apoptosis induced by anti-cancer drugs, but also to the study of apoptosis-related diseases in a living system.

The labelling of annexin V with indocyanine green (ICG) and π-conjugation extended ICG (ICG-C11) resulted in SWIR emitting probes that enable high-contrast molecular imaging of tumor apoptosis in living mice.     

Algorithm for mapping cutaneous tissue oxygen concentration using hyperspectral imaging

The measurement of tissue oxygenation plays an important role in the diagnosis and therapeutic assessment of a large variety of diseases. Many different methods have been developed and are currently applied in clinical practice for the measurement of tissue oxygenation. Unfortunately, each of these methods has its own limitations. In this paper we proposed the use of hyperspectral imaging as new method for the assessment of the tissue oxygenation level. To extract this information from hyperspectral images a new algorithm for mapping cutaneous tissue oxygen concentration was developed. This algorithm takes into account and solves some problems related to setting and calculation of some parameters derived from hyperspectral images. The algorithm was tested with good results on synthetic images and then validated on the fingers of a hand with different blood irrigation states. The results obtained have proved the ability of hyperspectral imaging together with the developed algorithm to map the oxy- and deoxyhemoglobin distribution on the analyzed fingers. These are only preliminary results and other studies should be done before this approach to be used in the clinical setting for the diagnosis and monitoring of various diseases.OCIS codes: (000.2170) Equipment and techniques, (110.4234) Multispectral and hyperspectral imaging, (170.6935) Tissue characterization, (300.6550) Spectroscopy, visible     

Variety identification of oat seeds using hyperspectral imaging: investigating the representation ability of deep convolutional neural network

Variety identification of seeds is critical for assessing variety purity and ensuring crop yield. In this paper, a novel method based on hyperspectral imaging (HSI) and deep convolutional neural network (DCNN) was proposed to discriminate the varieties of oat seeds. The representation ability of DCNN was also investigated. The hyperspectral images with a spectral range of 874–1734 nm were primarily processed by principal component analysis (PCA) for exploratory visual distinguishing. Then a DCNN trained in an end-to-end manner was developed. The deep spectral features automatically learnt by DCNN were extracted and combined with traditional classifiers (logistic regression (LR), support vector machine with RBF kernel (RBF_SVM) and linear kernel (LINEAR_SVM)) to construct discriminant models. Contrast models were built based on the traditional classifiers using full wavelengths and optimal wavelengths selected by the second derivative (2nd derivative) method. The comparison results showed that all DCNN-based models outperformed the contrast models. DCNN trained in an end-to-end manner achieved the highest accuracy of 99.19% on the testing set, which was finally employed to visualize the variety classification. The results demonstrated that the deep spectral features with outstanding representation ability enabled HSI together with DCNN to be a reliable tool for rapid and accurate variety identification, which would help to develop an on-line system for quality detection of oat seeds as well as other grain seeds.

The excellent representation ability of deep spectral features enables hyperspectral imaging combined with deep convolutional neural network to be a powerful tool for large-scale seeds detection in modern seed industry.     

Quantifying chlorophyll-a and b content in tea leaves using hyperspectral reflectance and deep learning

ABSTRACT

To improve the quality of green tea, low light stress has been used to increase the chlorophyll-a (chl-a) content of tea leaves, although shading treatments sometimes lead to early mortality of tea trees. Therefore, in situ measurement of chl-a and chlorophyll-b (chl-b), which are markers for evaluating light stress and response to changing environmental conditions, can be used to improve tea tree management. Chlorophyll content estimation is one of the most common applications of hyperspectral remote sensing, but most prior studies have evaluated samples grown under relatively low stress. Therefore, the results of prior studies are not applicable for estimating chl-a and chl-b contents of shade-grown tea. Machine learning algorithms have recently attracted attention as an approach for evaluating biochemical properties. In the present study, three different common machine learning algorithms were compared, including random forests, support vector machines and deep belief nets. The ratios of performance to deviation (RPDs) of deep belief nets (DBN) were always larger than 1.4 (the ranges of RPD were 1.49–4.92 and 1.48–5.10 for chl-a and chl-b, respectively), suggesting that DBN is a unique algorithm that can reliably be used for estimation of chl-a and chl-b contents.     

Application of hyperspectral imaging and chemometrics for variety classification of maize seeds

Seed variety classification is important for assessing variety purity and increasing crop yield. A hyperspectral imaging system covering the spectral range of 874–1734 nm was applied for variety classification of maize seeds. A total of 12 900 maize seeds including 3 different varieties were evaluated. Spectral data of 975.01–1645.82 nm were extracted and preprocessed. Discriminant models were developed using a radial basis function neural network (RBFNN). The influence of calibration sample size on classification accuracy was studied. Results showed that with the expansion of calibration sample size, calibration accuracy varied slightly, but prediction accuracy changed from the increasing form to the stable form. Accordingly, the optimal size of the calibration set was determined. Optimal wavelength selection was conducted by loading of principal components (PCs). The RBFNN model developed on optimal wavelengths with the optimal size of the calibration set obtained satisfactory results, with calibration accuracy of 93.85% and prediction accuracy of 91.00%. Visualization of classification map of seed varieties was achieved by applying this RBFNN model on the average spectra of each sample. Besides, the procedure to determine the optimal sample quantity proposed in this study was verified by support vector machine (SVM). The overall results indicated that hyperspectral imaging was a potential technique for variety classification of maize seeds, and would help to develop a real-time detection system for maize seeds as well as other crop seeds.

Hyperspectral imaging provides an effective way for seed variety classification for assessing variety purity and increasing crop yield.     

Application of hyperspectral imaging for spatial prediction of soluble solid content in sweet potato

Visible and near infrared (Vis-NIR) hyperspectral imaging was used for fast detection and visualization of soluble solid content (SSC) in ‘Beijing 553’ and ‘Red Banana’ sweet potatoes. Hyperspectral images were acquired from 420 ROIs of each cultivar of sliced sweet potatoes. There were 8 and 10 outliers removed from ‘Beijing 553’ and ‘Red Banana’ sweet potatoes by Monte Carlo partial least squares (MCPLS). The optimal spectral pretreatments were determined to enhance the performance of the prediction model. Successive projections algorithm (SPA) and competitive adaptive reweighted sampling (CARS) were employed to select characteristic wavelengths. SSC prediction models were developed using partial least squares regression (PLSR), support vector regression (SVR) and multivariate linear regression (MLR). The more effective prediction performances emerged from the SPA–SVR model with R_p² of 0.8581, RMSEP of 0.2951 and RPD_p of 2.56 for ‘Beijing 553’ sweet potato, and the CARS–MLR model with R_p² of 0.8153, RMSEP of 0.2744 and RPD_p of 2.09 for ‘Red Banana’ sweet potato. Spatial distribution maps of SSC were obtained in a pixel-wise manner using SPA–SVR and CARS–MLR models for quantifying the SSC level in a simple way. The overall results illustrated that Vis-NIR hyperspectral imaging was a powerful tool for spatial prediction of SSC in sweet potatoes.

Visible and near infrared (Vis-NIR) hyperspectral imaging was used for fast detection and visualization of soluble solid content (SSC) in ‘Beijing 553’ and ‘Red Banana’ sweet potatoes.     

Deeply-recursive convolutional neural network for Raman spectra identification

Raman spectroscopy has been widely used in various fields due to its unique and superior properties. For achieving high spectral identification speeds and high accuracy, machine learning methods have found many applications in this area, with convolutional neural network-based methods showing great advantages. In this study, we propose a Raman spectral identification method using a deeply-recursive convolutional neural network (DRCNN). It has a very deep network structure (up to 16 layers) for improving performance without introducing more parameters for recursive layers, which eases the difficulty of training. We also propose a recursive-supervision extension to ease the difficulty of training. By testing several different open-source spectral databases, DRCNN has achieved higher prediction accuracies and better performance in transfer learning compared with other CNN-based methods. Superior identification performance is demonstrated, especially by identification, for characteristically similar and indistinguishable spectra.

Raman spectroscopy has been widely used in various fields due to its unique and superior properties.     

An advanced method for mineral mapping applicable to hyperspectral images: the composite MSAM

We developed a new method for mineral mapping based on the continuum-removal Modified Spectral Angle Mapper (MSAM) method to distinguish minerals, such as calcite and chlorite, which have diagnostic spectral features in the visible near infrared (VNIR) and shortwave infrared (SWIR) regions. This method was applied to the reflectance data obtained by HyMap, a hyperspectral airborne sensor with 126 spectral bands (0.44–2.48 μm) in Cuprite, Nevada, USA. Reflectance spectra in the VNIR and SWIR regions are characterized by a convex background (continuum). Continuum-removal processing results in emphasis of the shape and location of absorption peaks in reflectance spectra, and thus in recognition of the chemical species which are causing the absorption spectrum. On the other hand, the broad features in continuum spectra contain essential information for mineral classification as well. Therefore, we produced mineral index maps from both the continuum and continuum-removed images based on the MSAM method and then combined them to generate a mineral distribution map. This composite MSAM method was applied to remotely sensed hyperspectral data for the first time. It was confirmed that our mineral index maps were consistent with the existing mineral maps. We can conclude that the composite MSAM method has a great advantage over our previously developed method in discrimination between calcite and chlorite, in particular.     

Multispectral near-IR reflectance and transillumination imaging of teeth

NIR imaging methods do not require ionizing radiation and have great potential for detecting caries lesions (tooth decay) on high-risk proximal and occlusal tooth surfaces and at the earliest stages of development. Previous in vitro and in vivo studies at 1300-nm demonstrated that high contrast reflectance and transillumination images could be acquired of caries lesions on tooth proximal and occlusal surfaces where most new decay is found. Water absorption varies markedly between 1200 and 1600-nm and the scattering properties of enamel and the underlying dentin have not been characterized in this region. Hyperspectral reflectance studies show lower reflectivity from sound enamel and dentin at NIR wavelengths with higher water absorption. The purpose of this imaging study was to determine which NIR wavelengths between 1200 and 1600-nm provide the highest contrast of demineralization or caries lesions for each of the different modes of NIR imaging, including transillumination of proximal and occlusal surfaces along with cross polarization reflectance measurements. A tungsten halogen lamp with several spectral filters and a Ge-enhanced CMOS focal plane array (FPA) sensitive from 400 to 1600-nm were used to acquire the images of caries lesions on extracted teeth. Artificial interproximal lesions were created on twelve tooth sections of 5 & 6-mm thickness that were used for transillumination imaging. Fifty-four extracted teeth with suspected occlusal lesions were also examined in both occlusal transillumination and reflectance imaging modes. Cavity preparations were also cut into whole teeth and filled with composite and used to compare the contrast between composite and enamel at NIR wavelengths. NIR wavelengths longer than 1400-nm are likely to have better performance for the transillumination of occlusal caries lesions while 1300-nm appears best for the transillumination of proximal surfaces. Loss of mobile water in enamel markedly reduced the transparency of the enamel at all NIR wavelengths. Significantly higher contrast was attained for reflectance measurements at wavelengths that have higher water absorption, namely 1460-nm. Wavelengths with higher water absorption also provided higher contrast of composite restorations.     

On-line automatic slice positioning for brain MR imaging

In clinical brain MR imaging protocols, the technician collects a quick localizer and manually positions the subsequent scans using the localizer as a guide. We present a method for automatic slice positioning using a rapidly acquired 3D localizer. The localizer is automatically aligned to a statistical atlas representing 40 healthy subjects. The atlas contains the probability of a given tissue type occurring at a given location in atlas space and the conditional probability distribution of the multi-spectral MRI intensity values for a given tissue class. Accurate rigid alignment of each subject to an atlas ensures that all patients' scans are acquired in a consistent manner. A further benefit is that slices are positioned consistently over time, so that scans of patients returning for follow-up imaging can be compared side-by-side to accurately monitor the progression of illness. The procedure also helps ensure that left/right asymmetries reflect true anatomy rather than being the result of oblique slice positioning relative to the underlying anatomy. The use of an atlas-based procedure eliminates the need to refer to a database of previously scanned images of the same patient and ensures corresponding alignment across scanners and sites, without requiring fiducial markers. Since the registration method is probabilistic, the registration error tends to increase smoothly in the presence of increasing noise and unusual anatomy or pathology rather than failing catastrophically. Translations and rotations relative to the atlas can be set so that planning can be done in anatomical space, rather than scanner coordinates, and stored as part of the protocol allowing standardization of slice orientations.     

Comparison of the reflectance method (Reflotron reflectance photometer) with the absorbance method (automatic analysers) for the determination of cholesterol

The European Atherosclerosis Society (1) and the Expert Panel of the US National Cholesterol Education Program (2) have issued detailed guide values for recognition and management of hyperlipidaemia in adults. In these guidelines, the diagnosis of dyslipidaemia based on the measurements of total cholesterol, triacylglycerols, HDL and LDL cholesterol plays an important role. A prerequisite for the desired success of interventive measures is the reliability of the analytical data. The aim of this study was to investigate the precision and accuracy of Reflotron Cholesterol, a method based on the dry chemistry principle. Accuracy was assessed by establishing the correlation with the standardized automated methods used in routine lipid diagnosis. In addition, it was also examined whether the Reflotron Cholesterol results in plasma and blood are comparable. The Reflotron cholesterol (sample: blood) showed a good correlation with the CHOD/PAP method on a Hitachi 737 instrument (sample: plasma). The median value of the differences of the test results was -0.4%. Similarly, the method comparison of Reflotron Cholesterol (sample: blood) versus CHOD/PAP method on a SMAC instrument (sample: plasma) showed that Reflotron produces slightly (1.8%) higher results. The Reflotron Cholesterol values obtained from blood samples were slightly lower than those from plasma samples (median value of the differences: -2.2%). The results suggest that for routine purposes Reflotron Cholesterol provides results which are in good agreement with those obtained by standardized wet chemistry methods.     

Near-IR fluorescence and reflectance confocal microscopy for imaging of quantum dots in mammalian skin

Understanding the skin penetration of nanoparticles (NPs) is an important concern due to the increasing presence of NPs in consumer products, including cosmetics. Technical challenges have slowed progress in evaluating skin barrier and NP factors that contribute to skin penetration risk. To limit sampling error and other problems associated with histological processing, many researchers are implementing whole tissue confocal or multiphoton microscopies. This work introduces a fluorescence and reflectance confocal microscopy system that utilizes near-IR excitation and emission to detect near-IR lead sulfide quantum dots (QDs) through ex vivo human epidermis. We provide a detailed prediction and experimental analysis of QD detection sensitivity and demonstrate detection of QD skin penetration in a barrier disrupted model. The unique properties of near-IR lead-based QDs will enable future studies that examine the impact of further barrier-disrupting agents on skin penetration of QDs and elucidate mechanistic insight into QD tissue interactions at the cellular level.