Non-Small Cell Lung Cancer(NSCLC) is the most common cause of cancer death in Canada and worldwide. As the immune component of the NSCLC tumor microenvironment(TME) is highly prognostic of patient outcome, further understanding of TME and the spatial organization of the immune cells within the TME is needed for better patient prognosis and treatment planning. Current immunohistochemistry techniques quantify immune cell counts and density, but generally cannot asses the spatial relationship between tumour and immune cells. We have developed a multiplexed Immunohistochemistry(mIHC) procedure combining multiple labels per round with several rounds, enabling analysis of immune cell populations on a slide through consecutive cycles of staining, destaining & hyperspectral imaging. By integrating serial imaging, sequential labeling & image registration, we are able to spatially map the TME. Robust, accurate, segmentation of cell nuclei for overlapping nuclei is one of the most significant unsolved issues in digital pathology. We have trained a deep learning segmentation method to accurately segment individual cell nuclei within overlapping clusters of nuclei. By combining a mIHC technique which enables the detection of multiple markers with deep learning segmentation methods to segment every individual cell nuclei in tissue sections with an accuracy comparable to human annotation, we can analyze the cell-cell interactions between immune and tumour cells, enhancing our ability to perform molecularly based single cell analysis of multiple cell types simultaneously within the tissue. These two techniques joined can be scaled up to the entire tissue section level, improving our understanding of the biological aggressiveness of NSCLC’s.
Significance: Optical scattering signals obtained from tissue constituents contain a wealth of structural information. Conventional intensity features, however, are mostly dictated by the overall morphology and mean refractive index of these constituents, making it very difficult to exclusively sense internal refractive index fluctuations.
Aim: We perform a systematic analysis to elucidate how changes in internal refractive index profile of cell nuclei can best be detected via optical scattering.
Approach: We construct stochastically inhomogeneous nuclear models and numerically simulate their azimuth-resolved scattering patterns. We then process these two-dimensional patterns with the goal of identifying features that directly point to subnuclear structure.
Results: Azimuth-dependent intensity variations over the side scattering range provide significant insights into subnuclear refractive index profile. A particular feature we refer to as contrast ratio is observed to be highly sensitive to the length scale and extent of refractive index fluctuations; further, this feature is not susceptible to changes in the overall size and mean refractive index of nuclei, thereby allowing for selective tracking of subnuclear structure that can be linked to chromatin distribution.
Conclusions: Our analysis will potentially pave the way for scattering-based assessment of chromatin reorganization that is considered to be a key hallmark of precancer progression.
We construct stochastically inhomogeneous epithelial cell models via simulation of Gaussian random fields; the extent and correlation length of subnuclear refractive index fluctuations are based on values quantified from high-resolution images of cervical tissue. We then employ the finite-difference time-domain method to simulate azimuth-resolved light scattering patterns of the constructed models. We process these two-dimensional patterns and calculate a series of Haralick features with the ultimate goal of identifying signatures that directly point to changes in subnuclear refractive index profile. Our results show that azimuthal contrast calculated over specific angular ranges is highly sensitive to the extent and correlation length of refractive index fluctuations. This metric is insensitive to changes in the overall size and mean refractive index of the constructed models, thereby allowing for selective tracking of changes in subnuclear refractive index variations.
Dual-mode endomicroscopy is a diagnostic tool for early cancer detection. It combines the high-resolution nuclear tissue contrast of fluorescence endomicroscopy with quantified depth-dependent epithelial backscattering as obtained by diffuse optical microscopy. In an in vivo pilot imaging study of 27 oral lesions from 21 patients, we demonstrate the complementary diagnostic value of both modalities and show correlations between grade of epithelial dysplasia and relative depth-dependent shifts in light backscattering. When combined, the two modalities provide diagnostic sensitivity to both moderate and severe epithelial dysplasia in vivo.
Colin Schlosser, Nico Bodenschatz, Sylvia Lam, Marette Lee, Jessica McAlpine, Dianne Miller, Dirk J. T. Van Niekerk, Michele Follen, Martial Guillaud, Calum MacAulay, Pierre Lane
Current diagnostic capabilities and limitations of fluorescence endomicroscopy in the cervix are assessed by qualitative and quantitative image analysis. Four cervical tissue types are investigated: normal columnar epithelium, normal and precancerous squamous epithelium, and stromal tissue. This study focuses on the perceived variability within and the subtle differences between the four tissue groups in the context of endomicroscopic in vivo pathology. Conclusions are drawn on the general ability to distinguish and diagnose tissue types, on the need for imaging depth control to enhance differentiation, and on the possible risks for diagnostic misinterpretations.
This paper addresses the problem of classifying cells expressing different biomarkers. A deep learning based method that can automatically localize and count the cells expressing each of the different biomarkers is proposed. To classify the cells, a Convolutional Neural Network (CNN) was employed. Images of Immunohistochemistry (IHC) stained slides that contain these cells were digitally scanned. The images were taken from digital scans of IHC stained cervical tissues, acquired for a clinical trial. More than 4,500 RGB images of cells were used to train the CNN. To evaluate our method, the cells were first manually labeled based on the expressing biomarkers. Then we performed the classification on 156 randomly selected images of cells that were not used in training the CNN. The accuracy of the classification was 92% in this preliminary data set. The results have shown that this method has a good potential in developing an automatic method for immunohistochemical analysis.
Reflectance confocal microscopy is successfully used in infant skin research. Infant skin structure, function, and composition are undergoing a maturation process. We aimed to uncover how the epidermal architecture and cellular topology change with time. Images were collected from three age groups of healthy infants between one and four years of age and adults. Cell centers were manually identified on the images at the stratum granulosum (SG) and stratum spinosum (SS) levels. Voronoi diagrams were used to calculate geometrical and topological parameters. Infant cell density is higher than that of adults and decreases with age. Projected cell area, cell perimeter, and average distance to the nearest neighbors increase with age but do so distinctly between the two layers. Structural entropy is different between the two strata, but remains constant with time. For all ages and layers, the distribution of the number of nearest neighbors is typical of a cooperator network architecture. The topological analysis provides evidence of the maturation process in infant skin. The differences between infant and adult are more pronounced in the SG than SS, while cell cooperation is evident in all cases of healthy skin examined.
Optical scattering provides an intrinsic contrast mechanism for the diagnosis of early precancerous changes in tissues. There have been a multitude of numerical studies targeted at delineating the relationship between cancer-related alterations in morphology and internal structure of cells and the resulting changes in their optical scattering properties. Despite these efforts, we still need to further our understanding of inherent scattering signatures that can be linked to precancer progression. As such, computational studies aimed at relating electromagnetic wave interactions to cellular and subcellular structural alterations are likely to provide a quantitative framework for a better assessment of the diagnostic content of optical signals. In this study, we aim to determine the influence of structural length-scale variations on two-dimensional light scattering properties of cells. We numerically construct cell models with different lower bounds on the size of refractive index heterogeneities and we employ the finite-difference time-domain method to compute their azimuth-resolved light scattering patterns. The results indicate that changes in length-scale variations can significantly alter the two-dimensional scattering patterns of cell models. More specifically, the degree of azimuthal asymmetry characterizing these patterns is observed to be highly dependent on the range of length-scale variations. Overall, the study described here is expected to offer useful insights into whether azimuth-resolved measurements can be explored for diagnostic purposes.
We examined and established the potential of ex-vivo confocal fluorescence microscopy for differentiating between normal cervical tissue, low grade Cervical Intraepithelial Neoplasia (CIN1), and high grade CIN (CIN2 and CIN3). Our objectives were to i) use Quantitative Tissue Phenotype (QTP) analysis to quantify nuclear and cellular morphology and tissue architecture in confocal microscopic images of fresh cervical biopsies and ii) determine the accuracy of high grade CIN detection via confocal microscopy. Cervical biopsy specimens of colposcopically normal and abnormal tissues obtained from 15 patients were evaluated by confocal fluorescence microscopy. Confocal images were analyzed and about 200 morphological and architectural features were calculated at the nuclear, cellular, and tissue level. For the purpose of this study, we used four features to delineate disease grade including nuclear size, cell density, estimated nuclear-cytoplasmic (ENC) ratio, and the average of three nearest Delaunay neighbors distance (3NDND). Our preliminary results showed ENC ratio and 3NDND correlated well with histopathological diagnosis. The Spearman correlation coefficient between each of these two features and the histopathological diagnosis was higher than the correlation coefficient between colposcopic appearance and histopathological diagnosis. Sensitivity and specificity of ENC ratio for detecting high grade CIN were both equal to 100%. QTP analysis of fluorescence confocal images shows the potential to discriminate high grade CIN from low grade CIN and normal tissues. This approach could be used to help clinicians identify HGSILs in clinical settings.
Dysplastic progression in epithelial tissues is linked to changes in morphology and internal structure of cell nuclei. These changes lead to alterations in nuclear light scattering profiles that can potentially be monitored for diagnostic purposes. Numerical tools allow for simulation of complex nuclear models and are particularly useful for quantifying the optical response of cell nuclei as dysplasia progresses. In this study, we first analyze a set of quantitative histopathology images from twenty cervical biopsy sections stained with Feulgen-thionin. Since Feulgen-thionin is stoichiometric for DNA, the images enable us to obtain detailed information on size, shape, and chromatin content of all the segmented nuclei. We use this extensive data set to construct realistic three-dimensional computational models of cervical cell nuclei that are representative of four diagnostic categories, namely normal or negative for dysplasia, mild dysplasia, moderate dysplasia, and severe dysplasia or carcinoma in situ (CIS). We then carry out finite-difference time-domain simulations to compute the light scattering response of the constructed models as a function of the polar scattering angle and the azimuthal scattering angle. The results show that these two-dimensional scattering patterns exhibit characteristic intensity ridges that change form with progression of dysplasia; pattern processing reveals that Haralick features can be used to distinguish moderately and severely dysplastic or CIS nuclei from normal and mildly dysplastic nuclei. Our numerical study also suggests that different angular ranges need to be considered separately to fully exploit the diagnostic potential of two-dimensional light scattering measurements.
New imaging technologies are changing the field of digital pathology. This field faces numerous challenges and there is a pressing need for standardization, calibration protocols, quality control and quantitative assessment. We have designed a new calibration imaging slide (Cancer Imaging Slide), specifically to measure the characteristics of old or new imaging systems or scanners. The layout of the slide consists of 138 boxes with the side length of 1.6 mm, containing objects of known morphologic and photometric characteristics. Among them, 112 boxes contain different permutations of circles, ovals, and squares. The circles have different radii, radius/pitch ratios and step transmissions. The ovals have different sizes and orientations. The squares are consistent in size and orientation but have different step transmission values. Also, 16 boxes contain three resolution test targets: crosses, USAF target and Siemens star. The last 10 boxes are blank boxes with different transmission values. Four slides were scanned and imaged on one commercial whole-slide scanner and one high resolution imaging system. After segmenting the images, about 200 features (photometric, morphologic and architectural) were measured with our in-house image processing software. The objective of the project is to develop a statistical process control using this new slide. In this paper, we describe the characteristics of the slide and present our preliminary results.
Dysplastic progression is known to be associated with changes in morphology and internal structure of cells. A detailed
assessment of the influence of these changes on cellular scattering response is needed to develop and optimize optical
diagnostic techniques. In this study, we first analyzed a set of quantitative histopathologic images from cervical biopsies
and we obtained detailed information on morphometric and photometric features of segmented epithelial cell nuclei.
Morphometric parameters included average size and eccentricity of the best-fit ellipse. Photometric parameters included
optical density measures that can be related to dielectric properties and texture characteristics of the nuclei. These
features enabled us to construct realistic three-dimensional computational models of basal, parabasal, intermediate, and
superficial cell nuclei that were representative of four diagnostic categories, namely normal (or negative for dysplasia),
mild dysplasia, moderate dysplasia, and severe dysplasia or carcinoma in situ. We then employed the finite-difference
time-domain method, a popular numerical tool in electromagnetics, to compute the angle-resolved light scattering
properties of these representative models. Results indicated that a high degree of variability can characterize a given
diagnostic category, but scattering from moderately and severely dysplastic or cancerous nuclei was generally observed
to be stronger compared to scattering from normal and mildly dysplastic nuclei. Simulation results also pointed to
significant intensity level variations among different epithelial depths. This suggests that intensity changes associated
with dysplastic progression need to be analyzed in a depth-dependent manner.
The successful management of oral cancer depends upon early detection, which relies heavily on the clinician's ability to discriminate sometimes subtle alterations of the infrequent premalignant lesions from the more common reactive and inflammatory conditions in the oral mucosa. Even among experienced oral specialists this can be challenging, particularly when using new wide field-of-view direct fluorescence visualization devices clinically introduced for the recognition of at-risk tissue. The objective of this study is to examine if quantitative cytometric analysis of oral brushing samples could facilitate the assessment of the risk of visually ambiguous lesions. About 369 cytological samples were collected and analyzed: (1) 148 samples from pathology-proven sites of SCC, carcinoma in situ or severe dysplasia; (2) 77 samples from sites with inflammation, infection, or trauma, and (3) 144 samples from normal sites. These were randomly separated into training and test sets. The best algorithm correctly recognized 92.5% of the normal samples, 89.4% of the abnormal samples, 86.2% of the confounders in the training set as well as 100% of the normal samples, and 94.4% of the abnormal samples in the test set. These data suggest that quantitative cytology could reduce by more than 85% the number of visually suspect lesions requiring further assessment by biopsy.
(Partial Abstract)
Confocal microscopy can provide real-time, 2-D and 3-D images of the cellular morphology and tissue architecture features that pathologists use to detect precancerous lesions without the need for tissue removal, sectioning, and staining. The utility of 3-D confocal image stacks of epithelial tissue for detecting dysplasia has not yet been explored. We aim to extract morphometry and tissue architecture information from 2-D confocal reflectance images and 3-D image stacks from fresh, unstained cervical biopsies and compare their potential for detecting dysplasia. Nine biopsies are obtained from eight patients; confocal images are acquired pre- and postacetic acid at multiple epithelial depths in 1.5 µm-intervals. Postacetic acid images are processed to segment cell nuclei; after segmentation, 2-D images taken at 50 um below the tissue surface, and the entire 3-D image stacks are processed to extract morphological and architectural features. Data are analyzed to determine which features gave the best separation between normal and high-grade cervical precancer. Most significant differences are obtained from parameters extracted from the 3-D image stacks. However, in all cases where the 2-D features were multiplicatively scaled by the depth of acquisition divided by the epithelial thickness or scaled by the scattering coefficient, the significance level is equal to or greater than the comparable feature extracted from the 3-D image stacks. A linear discriminant function previously developed to separate 19 samples of normal tissue and high-grade cervical precancer based on the nuclear-to-cytoplasm (N/C) ratio and epithelial scattering coefficient is prospectively applied to the nine biopsies examined to determine the accuracy with which it could separate normal tissue from cervical intra epithelial neoplasia (CIN) 2/3. For the entire data set of 28 biopsies, a sensitivity and specificity of 100% is produced using this discriminant function;
A series of hyperspectral transmission images of hematoxylin and eosin stained tissue sections from cervical biopsies
were acquired at 10 nm intervals and assembled into a hyperspectral image cube. Custom software providing extraction
of spectra at each pixel allows selection of images with maximum contrast for determination of selected features and
differentiation of tissue features. Illumination profiles were created using a spectrally and temporally programmable
light engine based on a spatial light modulator that can dynamically create any narrow or broadband spectral profile was
used to select illumination wavelengths. Images were acquired with a monochrome CCD camera. Several methods of
combining images from individual or composite spectral bands to recreate color images for pathologist review are
shown. Unlike current "mechanical" illumination systems employing optical filters, filter wheels, motors, shutters and
multiple control interfaces, the light engine integrates the lamp, wavelength control, intensity control and exposure
control in a simple MEMS based system, where the only moving part is the lamp cooling fan. Illumination can now be
programmed dynamically with digital control of all illumination parameters allowing wavelengths and intensities to be
changed much faster than with filter wheels, and providing exposure control orders of magnitude more precise than
mechanical shutters. This system can be integrated with digital imaging systems. Digitally controlled illumination is bit
additive with image data providing high dynamic range imaging with monochrome or with color imaging devices.
Performance of image analysis software for nuclear morphometric and tissue architecture analysis are compared for
different wavelength regions.
Considerable variation exists among pathologist in the interpretation of intraepithelial neoplasia making it difficult to determine the natural history of these lesion and to establish management guidelines for chemoprevention. The aim of the study is to evaluate architectural features of pre-neoplastic progression in lung cancer, and to search for a correlation between architectural index and conventional pathology. Quantitative architectural analysis was performed on a series of normal lung biopsies and Carcinoma In Situ (CIS). Centers of gravity of the nuclei within a pre-defined region of interest were used as seeds to generate a Voronoi Diagram. About 30 features derived from the Voronoi diagram, its dual the Delaunay tessellation, and the Minimum Spanning Tree were extracted. A discriminant analysis was performed to separate between the two groups. The architectural Index was calculated for each of the bronchial biopsies that were interpreted as hyperplasia, metaplasia, mild, moderate or severe dysplasia by conventional histopathology criteria. As a group, lesions classified as CIS by conventional histopathology criteria could be distinguished from dysplasia using the architectural Index. Metaplasia was distinct from hyperplasia and hyperplasia from normal. There was overlap between severe and moderate dysplasia but mild dysplasia could be distinguished form moderate dysplasia. Bronchial intraepithelial neoplastic lesions can be degraded objectively by architectural features. Combination of architectural features and nuclear morphometric features may improve the quantitation of the changes occurring during the intra-epithelial neoplastic process.
Given the demographics of current and ex-smoking populations in North America, lung cancer will be a major problem in the foreseeable future. Early detection and treatment of lung cancer holds great promise for the management of this disease. Unlike cervical cancer, the physical, complete removal/destruction of all dysplastic lesions in the bronchial tree is not possible; however, treatment of the lesions using a chemopreventive agent is. Intermediate biomarkers have been used to screen promising chemopreventive agents for larger population studies. We have examined the natural history of lung cancer development by following a group of subjects at high risk of developing lung cancer using fluorescence endoscopy to identify the areas of abnormality for biopsy. Approximately 900 biopsies have been collected in this fashion and graded by at least two experienced, expert pathologists. Using an interactive version of the Cyto-Savant (Oncometrics Imaging Corp.), cytometric and tissue architectural data were collected from these biopsies. Using only the data from the normal and invasive cancer biopsies, quantitative morphometric and architectural indices were generated and calculated for all the collected biopsies. These indices were compared with Loss of Heterozygosity (LOH) of ten sites commonly associated with cancer. These results and the application of these quantitative measures to two small chemoprevention studies will be reported.
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.