No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
High throughput assessment of cells and tissues: Bayesian classification of spectral metrix from infrared vibrational spectroscopic imaging data
Abstract
Vibrational spectroscopy allows a visualization of tissue constituents based on intrinsic chemical composition and provides a potential route to obtaining diagnostic markers of diseases. Characterizations utilizing infrared vibrational spectroscopy, in particular, are conventionally low throughput in data acquisition, generally lacking in spatial resolution with the resulting data requiring intensive numerical computations to extract information. These factors impair the ability of infrared spectroscopic measurements to represent accurately the spatial heterogeneity in tissue, to incorporate robustly the diversity introduced by patient cohorts or preparative artifacts and to validate developed protocols in large population studies. In this manuscript, we demonstrate a combination of Fourier transform infrared (FTIR) spectroscopic imaging, tissue microarrays (TMAs) and fast numerical analysis as a paradigm for the rapid analysis, development and validation of high throughput spectroscopic characterization protocols. We provide an extended description of the data treatment algorithm and a discussion of various factors that may influence decision-making using this approach. Finally, a number of prostate tissue biopsies, arranged in an array modality, are employed to examine the efficacy of this approach in histologic recognition of epithelial cell polarization in patients displaying a variety of normal, malignant and hyperplastic conditions. An index of epithelial cell polarization, derived from a combined spectral and morphological analysis, is determined to be a potentially useful diagnostic marker.
... Phenotypical changes, including changes in metabolites, nuclear morphology, and nano-architecture, are more consistent across patients than the myriad of individual mutations and disrupted pathways underlying the disease, and can potentially better characterize tumors. This approach has shown very promising results for early cancer detection [21][22][23][24][25]28,29 , and-to a more limited extent-assessing cancer aggressiveness 26,27,30 . ...
... Furthermore, other optical technologies have shown very promising results to help improve histopathology, using methods such as UV excited florescence/auto-florescence 61,73,74 , infrared 21,29,75 and Raman Scattering 76,77 . However, there are important limitations associated with each approach. ...
... Next, we used 54 regions from 10 patients that contained representative biologically structures in prostate tissue, as the training data-set for our model (~13.5 billion spectra). The remaining regions (21) from the other 5 patients were used as the testing data set to evaluate the color transformation model. The important point about the testing data set is that the regions come from completely independent patients and no regions from testing patients are used in the training process. ...
Identifying prostate cancer patients that are harboring aggressive forms of prostate cancer remains a significant clinical challenge. Here we develop an approach based on multispectral deep-ultraviolet (UV) microscopy that provides novel quantitative insight into the aggressiveness and grade of this disease, thus providing a new tool to help address this important challenge. We find that UV spectral signatures from endogenous molecules give rise to a phenotypical continuum that provides unique structural insight (i.e., molecular maps or “optical stains") of thin tissue sections with subcellular (nanoscale) resolution. We show that this phenotypical continuum can also be applied as a surrogate biomarker of prostate cancer malignancy, where patients with the most aggressive tumors show a ubiquitous glandular phenotypical shift. In addition to providing several novel “optical stains” with contrast for disease, we also adapt a two-part Cycle-consistent Generative Adversarial Network to translate the label-free deep-UV images into virtual hematoxylin and eosin (H&E) stained images, thus providing multiple stains (including the gold-standard H&E) from the same unlabeled specimen. Agreement between the virtual H&E images and the H&E-stained tissue sections is evaluated by a panel of pathologists who find that the two modalities are in excellent agreement. This work has significant implications towards improving our ability to objectively quantify prostate cancer grade and aggressiveness, thus improving the management and clinical outcomes of prostate cancer patients. This same approach can also be applied broadly in other tumor types to achieve low-cost, stain-free, quantitative histopathological analysis.
... The poor predictive power may be attributed to the vast genetic heterogeneity of tumors, which makes it extremely difficult to identify a unique set of mutations that provide reliable prognostic information. Alternatively, recent efforts have shifted towards exploring phenotypical "common-denominators" to the countless genetic and epigenetic alterations that lead to cancer [13][14][15][16][17][18][19][20][21][22][23][24][25][26][27][28][29]. Phenotypical changes, including changes in metabolites, nuclear morphology, and nano-architecture, are more consistent across patients than the myriad of individual mutations and disrupted pathways underlying the disease, and can potentially better characterize tumors. ...
... Phenotypical changes, including changes in metabolites, nuclear morphology, and nano-architecture, are more consistent across patients than the myriad of individual mutations and disrupted pathways underlying the disease, and can potentially better characterize tumors. This approach has shown very promising results for early cancer detection [13][14][15][16][17][18][19][20][21][22][23][24][25][26], and-to a more limited extent-assessing cancer aggressiveness [27][28][29]. ...
... Indeed other optical technologies have shown very promising results to help improve histopathology, using methods such as UV excited florescence/auto-florescence [60,71,72], infrared [13,16,18] and Raman Scattering [73,74]. However, there are important limitations associated with each approach. ...
Identifying prostate cancer patients that are harboring aggressive forms of prostate cancer remains a significant clinical challenge. Here we develop an approach based on multispectral deep-ultraviolet (UV) microscopy that provides novel quantitative insight into the aggressiveness and grade of this disease, thus providing a new tool to help address this important challenge. We find that UV spectral signatures from endogenous molecules give rise to a phenotypical continuum that provides unique structural insight (i.e., molecular maps or “optical stains") of thin tissue sections with subcellular (nanoscale) resolution. We show that this phenotypical continuum can also be applied as a surrogate biomarker of prostate cancer malignancy, where patients with the most aggressive tumors show a ubiquitous glandular phenotypical shift. In addition to providing several novel “optical stains” with contrast for disease, we also adapt a two-part Cycle-consistent Generative Adversarial Network to translate the label-free deep-UV images into virtual hematoxylin and eosin (H&E) stained images, thus providing multiple stains (including the gold-standard H&E) from the same unlabeled specimen. Agreement between the virtual H&E images and the H&E-stained tissue sections is evaluated by a panel of pathologists who find that the two modalities are in excellent agreement. This work has significant implications towards improving our ability to objectively quantify prostate cancer grade and aggressiveness, thus improving the management and clinical outcomes of prostate cancer patients. This same approach can also be applied broadly in other tumor types to achieve low-cost, stain-free, quantitative histopathological analysis.
... Thus, there is a need for molecular based techniques to grade tissue samples in a reliable and reproducible manner. FTIR imaging of microarrays was coupled with statistical pattern recognition techniques in order to demonstrate histopathologic characterization of prostatic tissue and to differentiate benign from malignant prostatic epithelium [6][7][8][9][10]. ...
... 1 Immersion in hexane at 40°C for 48h and or in sequence of 3 − 4 baths [6,8,31]; 2 Immersion in xylene in sequence of 3 − 4 baths for 5 minutes, then wash and clean in acetone for 5 minutes, and leave to airdry [19,26]; 3 Wash in an orbital mixer with Citro clear for 6 or 20 minutes, and then acetone at 48°C for a further 6 or 20 minutes before air-drying for 1h under ambient conditions [11,27]. ...
... In general, mapping studies typically examine samples at coarse spatial resolution (>20 μm) and investigate small numbers of pa- CaF2 [20,32,33] Less reflective loss at the substrate-sample interface Biochemically compatible for cell growth BaF2 [6,8,22,24,25,34] Not suitable for cell growth due to low cell viability ZnSe [36] Higher refractive index could lead to strong interference between refracted light beams Not suitable for cell growth due to low cell viability Low-e Mirr IR infrared reflecting plates (Kevley technologies, Ohio, USA) [11,19,[22][23][24]27,29,30,32] Suitable for in situ cell culture Not suitable for samples where the thickness is undetermined or is less than the wavelength of the infrared light Spotlight 300 (Perkin Elmer Inc.) infrared imaging spectrometer. ...
This paper presents a retrospective study from 2004 to 2014 of FTIR prostate cancer spectroscopy related to tissues and cell biology. Since vibrational spectroscopy is delicately sensitive to the biochemical composition of the sample and variations therein, it is possible to monitor metabolic processes in tissue and cells, and to construct spectral maps based on thousands of collected IR spectra. These reveal information on tissue structure, distribution of cellular components, metabolic activity and the health condition of cells and tissues. In addition, rapid collection, reliable data, a powerful ability to structure elucidation about IR spectroscopy, and the need for a rapid diagnosis of traditional biopsy (subject to sampling and inter-observer) have potentiated infrared as a way for a new type of analysis based on optical examination and being more objective than conventional colour methods.
... There is considerable interest in the detection of cancer by applying machine learning algorithms to the analysis of the extensive datasets obtained by the application of infrared (IR) imaging spectroscopies to fixed human tissue. [1][2][3][4][5][6][7][8][9] Baker et al. 10,11 demonstrated considerable improvement in sensitivity and specificity in the Gleason grading of prostate cancer when applying principal component discriminant function analysis (PC-DFA) to a Fourier transform infrared (FTIR) imaging dataset. Similarly, the application of convolutional neural networks to a combination of results obtained from FTIR spectral imaging and associated spatial information obtained from tissue microarrays was able to identify six major cellular and acellular constituents associated with breast cancer. ...
... 3 There have been several reviews of advances in the instrumentation and application of the FTIR technique to cancer [12][13][14][15] and the application of techniques for obtaining chemical information from FTIR. 3,[6][7][8][9] We recently applied a novel machine learning multivariate metrics analysis (MA) technique to the analysis of FTIR images obtained from four cell lines associated with oesophageal cancer 16 and compared its performance with the well-established random forest (RF) method. The MA was found to achieve greater accuracy in discriminating between the cell types in a shorter time than the RF method. ...
A novel machine learning algorithm is shown to accurately discriminate between oral squamous cell carcinoma (OSCC) nodal metastases and surrounding lymphoid tissue on the basis of a single metric, the...
... The poor predictive power may be attributed to the vast genetic heterogeneity of tumors, which makes it extremely difficult to identify a unique set of mutations that provide reliable prognostic information. Alternatively, recent efforts have shifted towards exploring phenotypical "commondenominators" to the countless genetic and epigenetic alterations that lead to cancer (13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29). Phenotypical changes, including changes in metabolites, nuclear morphology, and nanoarchitecture, are more consistent across patients than the myriad of individual mutations and disrupted pathways underlying the disease, and can potentially better characterize tumors. ...
... Phenotypical changes, including changes in metabolites, nuclear morphology, and nanoarchitecture, are more consistent across patients than the myriad of individual mutations and disrupted pathways underlying the disease, and can potentially better characterize tumors. This approach has shown very promising results for early cancer detection (13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26), and-to a more limited extent-assessing cancer aggressiveness (27)(28)(29). ...
Identifying prostate cancer patients that are harboring aggressive forms of prostate cancer remains a significant clinical challenge. To shed light on this problem, we develop an approach based on multispectral deep-ultraviolet (UV) microscopy that provides novel quantitative insight into the aggressiveness and grade of this disease. First, we find that UV spectral signatures from endogenous molecules give rise to a phenotypical continuum that differentiates critical structures of thin tissue sections with subcellular spatial resolution, including nuclei, cytoplasm, stroma, basal cells, nerves, and inflammation. Further, we show that this phenotypical continuum can be applied as a surrogate biomarker of prostate cancer malignancy, where patients with the most aggressive tumors show a ubiquitous glandular phenotypical shift. Lastly, we adapt a two-part Cycle-consistent Generative Adversarial Network to translate the label-free deep-UV images into virtual hematoxylin and eosin (H&E) stained images. Agreement between the virtual H&E images and the gold standard H&E-stained tissue sections is evaluated by a panel of pathologists who find that the two modalities are in excellent agreement. This work has significant implications towards improving our ability to objectively quantify prostate cancer grade and aggressiveness, thus improving the management and clinical outcomes of prostate cancer patients. This same approach can also be applied broadly in other tumor types to achieve low-cost, stain-free, quantitative histopathological analysis.
... Machine learning is frequently applied to spectroscopy and spectroscopic imaging to differentiate distinct tissue types [52][53][54][55][56][57][58][59][60][61][62][63][64][65][66][67][68][69], while recent methods attempt to map molecular spectra to conventional stains [70,71] to produce images that can be interpreted by pathologists without additional training. ...
... Most methods of FTIR classification leverage only individual pixels (spectra). Many traditional unsupervised and supervised approaches have been applied to FTIR data, including as k-means clustering [98], hierarchical cluster analysis (HCA) [99], Bayesian classifiers [55,66,100], random forest classifiers [63,66,67], support vector machines (SVMs) [66] and linear discriminant analysis [101], and ANNs [32,66,102]. Deep learning has not been fully explored due to the lack of spatial detail introduced by most FTIR imaging instruments, however recent advances in instrument resolution allows the application of CNNs [69,71]. ...
Deep neural networks are emerging as a popular choice for hyperspectral image analysis—compared with other machine learning approaches, they are more effective for a variety of applications in hyperspectral imaging. Part I (Chap. 3) introduces the fundamentals of deep learning algorithms and techniques deployed with hyperspectral images. In this chapter (Part II), we focus on application-specific nuances and design choices with respect to deploying such networks for robust analysis of hyperspectral images. We provide quantitative and qualitative results with a variety of deep learning architectures, and compare their performance to baseline state-of-the-art methods for both remote sensing and biomedical image analysis tasks. In addition to surveying recent developments in these areas, our goal in these two chapters is to provide guidance on how to utilize such algorithms for multichannel optical imagery. With that goal, we also provide code and example datasets used in this chapter.
... The most common approach leverages annotated IR images to train supervised machine learning (ML) algorithms for automated classification. [29][30][31][32][33] This approach would require additional training for experts to integrate quantitative IR maps with existing clinical practice. For example, immunohistochemical labels frequently rely on counterstains to label tissue landmarks that are currently difficult to identify in IR images. ...
... K-means clustering 6,34 or hierarchical cluster analysis 34 and supervised techniques such as random forests, 33,35 Bayesian, 30,32 artificial neural network (ANN) 32,36 and support vector machines (SVMs) 32,37 . ...
Histological stains, such as hematoxylin and eosin (H&E), are routinely used in clinical diagnosis and research. While
these labels offer a high degree of specificity, throughput is limited by the need for multiple samples. Traditional
histology stains, such as immunohistochemical labels, also rely only on protein expression and cannot quantify small
molecules and metabolites that may aid in diagnosis. Finally, chemical stains and dyes permanently alter the tissue,
making downstream analysis impossible. Fourier transform infrared (FTIR) spectroscopic imaging has shown promise
for label-free characterization of important tissue phenotypes, and can bypass the need for many chemical labels.
FTIR classification commonly leverages supervised learning, requiring human annotation that is tedious and prone to
errors. One alternative is digital staining, which leverages machine learning to map infrared spectra to a corresponding
chemical stain. This replaces human annotation with computer-aided alignment. Previous work relies on alignment of
adjacent serial tissue sections. Since the tissue samples are not identical at the cellular level, this technique cannot be
applied to high-definition FTIR images. In this paper, we demonstrate that cellular-level mapping can be accomplished
using identical samples for both FTIR and chemical labels. In addition, higher-resolution results can be achieved using
a deep convolutional neural network that integrates spatial and spectral features.
... prin-191 cipal component analysis) and clustering (e.g. k-means or 192 hierarchical clustering) of spectral values at each pixel to im-193 prove the segmentation performance [14], [25]. However, such HSI images [14], [25], [26]. ...
... k-means or 192 hierarchical clustering) of spectral values at each pixel to im-193 prove the segmentation performance [14], [25]. However, such HSI images [14], [25], [26]. ...
Hyperspectral imaging (HSI) of tissue samples in the mid-infrared (mid-IR) range provides spectro-chemical and tissue structure information at sub-cellular spatial resolution. Disease-states can be directly assessed by analyzing the mid- IR spectra of different cell-types (e.g. epithelial cells) and sub cellular components (e.g. nuclei), provided we can accurately classify the pixels belonging to these components. The challenge is to extract information from hundreds of noisy mid-IR bands at each pixel, where each band is not very informative in itself, making annotations of unstained tissue HSI images particularly tricky. Because the tissue structure is not necessarily identical between the two sections, only a few regions in unstained HSI image can be annotated with high confidence, even when serial (or adjacent) H&E stained section is used as a visual guide. In order to completely use both labeled and unlabeled pixels in training images, we have developed an HSI pixel classification method that uses semi-supervised learning for both spectral dimension reduction and hierarchical pixel clustering. Compared to supervised classifiers, the proposed method was able to account for the vast differences in spectra of sub-cellular components of the same cell-type and achieve an F1-score of 71.18% on two fold cross-validation across 20 tissue images. To generate further interest in this promising modality we have released our source code and also showed that disease classification is straightforward after HSI image segmentation.
... In order to discriminate different species of Oceanobacillus, five machine learning models including k-nearest neighbour (kNN) (Zhang et al., 2018), logistic regression (LR) (Stoltzfus, 2011), random forest (RF) (Ho, 1998), support vector machine (SVM) (Tolstik et al., 2014) and Gaussian Naive Bayesian (GNB) (Bhargava et al., 2006) were established. The collected Raman spectra of six Oceanobacillus strains were used for model establishment, of which 70% were used for training and 30% for testing. ...
Many traditional fermented foods and beverages industries around the world request the addition of multi‐species starter cultures. However, the microbial community in starter cultures is subject to fluctuations due to their exposure to an open environment during fermentation. A rapid detection approach to identify the microbial composition of starter culture is essential to ensure the quality of the final products. Here, we applied single‐cell Raman spectroscopy (SCRS) combined with machine learning to monitor Oceanobacillus species in Daqu starter, which plays crucial roles in the process of Chinese baijiu. First, a total of six Oceanobacillus species (O. caeni, O. kimchii, O. iheyensis, O. sojae, O. oncorhynchi subsp. Oncorhynchi and O. profundus) were detected in 44 Daqu samples by amplicon sequencing and isolated by pure culture. Then, we created a reference database of these Oceanobacillus strains which correlated their taxonomic data and single‐cell Raman spectra (SCRS). Based on the SCRS dataset, five machine‐learning algorithms were used to classify Oceanobacillus strains, among which support vector machine (SVM) showed the highest rate of accuracy. For validation of SVM‐based model, we employed a synthetic microbial community composed of varying proportions of Oceanobacillus species and demonstrated a remarkable accuracy, with a mean error was less than 1% between the predicted result and the expected value. The relative abundance of six different Oceanobacillus species during Daqu fermentation was predicted within 60 min using this method, and the reliability of the method was proved by correlating the Raman spectrum with the amplicon sequencing profiles by partial least squares regression. Our study provides a rapid, non‐destructive and label‐free approach for rapid identification of Oceanobacillus species in Daqu starter culture, contributing to real‐time monitoring of fermentation process and ensuring high‐quality products.
... Although IR spectra are difficult to interpret by direct inspection, significant progress has been made by the application of machine learning techniques. [1][2][3][4][5][6] Both FTIR and Raman spectroscopy have been used to study oral squamous cell carcinoma (OSCC) (see Byrne et al. 7 for a review). In particular, Fukuyama et al. 8 reported that the FTIR spectra of normal tissue have stronger contributions from keratin and collagen than abnormal tissue. ...
A machine learning algorithm (MLA) has predicted the prognosis of oral potentially malignant lesions and discriminated between lymph node tissue and metastatic oral squamous cell carcinoma (OSCC). The MLA analyses metrics, which are ratios of Fourier transform infrared absorbances, and identifies key wavenumbers that can be associated with molecular biomarkers. The wider efficacy of the MLA is now shown in the more complex primary OSCC tumour setting, where it is able to identify seven types of tissue. Three epithelial and four non-epithelial tissue types were discriminated from each other with sensitivities between 82% and 96% and specificities between 90% and 99%. The wavenumbers involved in the five best discriminating metrics for each tissue type were tightly grouped, indicating that small changes in the spectral profiles of the different tissue types are important. The number of samples used in this study was small, but the information will provide a basis for further, larger investigations.
... www.nature.com/scientificreports/ cases relying on sophisticated machine learning algorithms for data analysis 26,30 . Similar applications have been demonstrated using Raman spectroscopy [31][32][33][34] . ...
Tissue microarrays (TMAs) are commonly used for the rapid analysis of large numbers of tissue samples, often in morphological assessments but increasingly in spectroscopic analysis, where specific molecular markers are targeted via immunostaining. Here we report the use of an automated high-throughput system based on desorption electrospray ionization (DESI) mass spectrometry (MS) for the rapid generation and online analysis of high-density (6144 samples/array) TMAs, at rates better than 1 sample/second. Direct open-air analysis of tissue samples (hundreds of nanograms) not subjected to prior preparation, plus the ability to provide molecular characterization by tandem mass spectrometry (MS/MS), make this experiment versatile and applicable to both targeted and untargeted analysis in a label-free manner. These capabilities are demonstrated in a proof-of-concept study of frozen brain tissue biopsies where we showcase (i) a targeted MS/MS application aimed at identification of isocitrate dehydrogenase mutation in glioma samples and (ii) an untargeted MS tissue type classification using lipid profiles and correlation with tumor cell percentage estimates from histopathology. The small sample sizes and large sample numbers accessible with this methodology make for a powerful analytical system that facilitates the identification of molecular markers for later use in intraoperative applications to guide precision surgeries and ultimately improve patient outcomes.
... Savitzky Golay 9 point smoothening, baseline correction, and normalization to amide I peak at 1650 cm −1 were done for each core in the TMA. The spectral data were converted to spectral metrics using methods described previously (56,57). Briefly, we calculated spectral metrics as ratios of peak heights, peak area and heights, peak areas, and centroid wave number locations of the peaks. ...
The structure and organization of a tumor and its microenvironment are often associated with cancer outcomes due to spatially varying molecular composition and signaling. A persistent challenge is to use this physical and chemical spatial organization to understand cancer progression. Here, we present a high-definition infrared imaging–based organizational measurement framework (INFORM) that leverages intrinsic chemical contrast of tissue to label unique components of the tumor and its microenvironment. Using objective and automated computational methods, further, we determine organization characteristics important for prediction. We show that the tumor spatial organization assessed with this framework is predictive of overall survival in colon cancer that adds to capability from clinical variables such as stage and grade, approximately doubling the risk of death in high-risk individuals. Our results open an all-digital avenue for measuring and studying the association between tumor spatial organization and disease progression.
... 33 The acquisition of a limited spectral range via a discrete subset of spectral positions, typically less than 30 bands, can often be used to reduce data acquisition time while maintaining the analytical capabilities of IR spectroscopy and imaging. 38 Mohamed et al. showed that 2D covariance analysis using 4 wavenumbers can identify the 4 main cell types present in breast cancer tissue sections. 39 They could accurately classify tissue sections recorded by transflectance in the presence of paraffin while the training set was obtained on dewaxed tissue sections by transmission, even when the test set was collected with a different brand of FTIR microscope and a different pixel size. ...
Molecular imaging has the potential to unearth important spatial and temporal relationships in biological systems, including intercellular signaling as well as environmentally-cued morphological changes. Various chemical imaging techniques show functional groups, molecular weights, etc., but no individual technique to date has the ability to simultaneously access all chemical information. Thus, it is highly attractive to combine information from two or more analytical techniques. Multimodal imaging is a recent, highly effective strategy for acquiring images by combining chemical information from multiplexed platforms.(1) This emerging integrated imaging approach yields information unattainable from a single method, enabling the evaluation of subtle biochemical changes and opening the way for a quantitative molecular overview of the morphological structure in biological tissues or architecture of cells. As a result, qualitative and quantitative multiomics investigations have the capability to revolutionize our understanding of disease progression and the healing process. Such a strategy has received increasing attention as it helps to elucidate the complex spatial distribution of biomolecules from the surface of a biological sample while circumventing the specific limitations of an individual imaging technique.
... We then use this spectral information to determine outcome-associated patient groups. Fourier transform infrared (FT-IR) spectroscopic imaging is effective in providing spatially specific molecular analysis of samples for pathology, without the need for stains or dyes [7][8][9][10][11] . It can be used to probe biomolecular changes in the tumor microenvironment [8][9][10]12 for comprehensive spatio-molecular assessment of the tissue which can be useful for determining prognosis 13 . ...
Molecular analysis techniques such as gene expression analysis and proteomics have contributed greatly to our understanding of cancer heterogeneity. In prior studies, gene expression analysis was shown to stratify patient outcome on the basis of tumor-microenvironment associated genes. A specific gene expression profile, referred to as ECM3 (Extracellular Matrix Cluster 3), indicated poorer survival in patients with grade III tumors. In this work, we aimed to visualize the downstream effects of this gene expression profile onto the tissue, thus providing a spatial context to altered gene expression profiles. Using infrared spectroscopic imaging, we identified spectral patterns specific to the ECM3 gene expression profile, achieving a high spectral classification performance of 0.87 as measured by the area under the curve of the receiver operating characteristic curve. On a patient level, we correctly identified 20 out of 22 ECM3 group patients and 19 out of 20 non-ECM3 group patients by using this spectroscopic imaging-based classifier. By comparing pixels that were identified as ECM3 or non-ECM3 with H&E and IHC images, we were also able to observe an association between tissue morphology and the gene expression clusters, showing the ability of our method to capture broad outcome associated features from infrared images.
... For a comprehensive non-destructive imaging platform, infrared microscopic imaging in the mid-infrared (MIR) region was added to detect and characterize molecular components. This analytical technique is based on the absorption of MIR radiation by vibrational transitions in covalent bonds and allows molecular imaging of biological specimens (Bhargava et al., 2006). Without time-consuming extraction, purification and separation steps, and by maintaining the topographic integrity, it is possible to acquire unique images of the spatial dispersion of carbohydrates, proteins, lipids, nucleic acids, cholesterols, phospholipids and other smaller molecules with high areal resolution (Fernandez et al., 2005), providing crucial chemo-morphological information (Steiner and Koch, 2009). ...
Information on the adaptation of bone structures during evolution is rare since histological data are limited. Micro- and nano-computed tomography of a fossilized vertebra from Champsosaurus sp., which has an estimated age of 70-73 million years, revealed lower porosity and higher bone density compared to modern Crocodylidae vertebrae. Mid-infrared reflectance and energy dispersive X-ray mapping excluded a petrification process, and demonstrated a typical carbonate apatite distribution, confirming histology in light- and electron microscopy of the preserved vertebra. As a consequence of this evolutionary process, the two vertebrae of modern Crocodylidae show reduced overall stiffness in the finite element analysis simulation compared to the fossilized Champsosaurus sp. vertebra, with predominant stiffness along the longitudinal z-axes.
... One advantage of this approach is that the results are independent of absolute absorbance and thus insensitive to factors such as sample thickness or normalization of the spectra. Importantly, this MA method treats all the data equally and does not attribute any biological significance to any particular wavenumber, in contrast to other work such as Fernandez et al. [32,33] in which discrimination of prostate tissues used metrics that were defined to have a significance related to tissue biochemistry. By examining ratios at wavenumbers over the whole range of 1000 cm −1 to 1800 cm −1 , the MA demonstrates the existence of biomarkers at wavenumbers that have not been identified in previous studies using other analysis techniques. ...
It is demonstrated that a novel multivariate analysis technique can discriminate with accuracies in the range 81–97% between Fourier transform infrared (FTIR) images of esophageal cancer OE19 and OE21 cell lines, and between esophageal cancer associated myofibroblast (CAM) and adjacent tissue myofibroblast (ATM) cells. The latter cells are morphologically indistinguishable but are known to have functionally important differences in their capacity to stimulate cancer cell growth; this report provides the first accurate spectral discrimination between CAM and ATM cells taken from the same patient. Rapid and accurate discrimination between cell types was achieved, and key wavenumbers were identified which uniquely discriminate between all four cell types. This metrics-based analysis (MA) method is shown to be unique for distinguishing between cancer stromal cells from the same patient. The key wavenumbers differ significantly from those typically found to discriminate between various esophageal cell and tissue types. A comparison is made between the MA and the established Random Forest method, and the advantages of the MA are discussed. Crucially the findings suggest a novel method that allows cancer staging based discrimination of the stromal cell types that provide the niche for tumor development.
... In most tissue segmentation studies, the entire spectrum is utilized for simultaneous analysis of chemical changes and alterations in tissue morphology. The idea of using discrete features was popularized (14) when data sets got very large and precluded full spectral analyses in reasonable times but increased significantly in importance with the advent of discrete frequency IR imaging (15,16). One direction has been focused on variable selection (17)(18)(19) to speed up biomedical imaging by reduced data acquisition (20,21). ...
Infrared (IR) spectroscopic imaging, utilizing both the molecular and structural disease signatures, enables extensive profiling of tumors and their microenvironments. Here, we examine the relative merits of using either the fingerprint or the high frequency regions of the IR spectrum for tissue histopathology. We selected a complex model as a test case, evaluating both stromal and epithelial segmentation for various breast pathologies. IR spectral classification in each of these spectral windows is quantitatively assessed by estimating area under the curve (AUC) of the receiver operating characteristic curve (ROC) for pixel level accuracy and images for diagnostic ability. We found only small differences, though some that may be sufficiently important in diagnostic tasks to be clinically significant, between the two regions with the fingerprint region-based classifiers consistently emerging as more accurate. The work provides added evidence and comparison with fingerprint region, complex models, and previously untested tissue type (breast) – that the use of restricted spectral regions can provide high accuracy. Our study indicates that the fingerprint region is ideal for epithelial and stromal models to obtain high pixel level accuracies. Glass slides provide a limited spectral feature set but provides accurate information at the patient level.
... 20,30 While these methods generally involve the full spectrum or a principal components transformed data set, another option is to focus on parts of the spectrum that are known to be responsive to concentrations of the desired species. For example, using carefully selected spectral features that are biochemically significant and low noise, efficient in classification models based on mid-infrared imaging 32 have shown success by using ratios of bands' properties (e.g. areas, heights, centers of gravity), usually termed as metrics, as the input for classification. ...
Rapid measurements of protein and oil content are important for a variety of uses, from sorting of soybeans at the point of harvest to feedback during soybean meal production. In this study, our goal is to develop a simple protocol to permit rapid and robust quantitative prediction of soybean constituents using transmission Raman spectroscopy (TRS). To develop this approach, we systematically varied the various elements of the measurement process to provide a diverse test bed. First, we utilized an in-house-built benchtop TRS instrument such that suitable optical configurations could be rapidly deployed and analyzed for experimental data collection for individual soybean grains. Second, we also utilized three different soybean varieties with relatively low (33.97%), medium (36.98%), and high protein (41.23%) contents to test the development process. Third, samples from each variety were prepared using whole bean and three different sample treatments (i.e., ground bean, whole meal, and ground meal). In each case, we modeled the data obtained using partial least squares (PLS) regression and assessed spectral metric-based multiple linear regression (metric-MLR) approaches to build robust prediction models. The metric-MLR models showed lower root mean square errors (RMSEPs), and hence better prediction, compared to corresponding classical PLS regression models for both bulk protein and oil for all treatment types. Comparing different sample preparation approaches, a lower RMSEPs was observed for whole meal treatment and thus the metric-MLR modeling with ground meal treatment was considered to be optimal protocol for bulk protein and oil prediction in soybean, with RMSEP values of 1.15 ± 0.04 (R²= 0.87) and 0.80 ± 0.02 (R²= 0.87) for bulk protein and oil, respectively. These predictions were nearly two- to threefold better (i.e., lower RMSEPs) than the corresponding NIR spectroscopy measurements (i.e., secondary gold standards in grain industry). For content prediction in whole soybean, incorporating physical attributes of individual grains in metric-MLR approach show up to 22% improvement in bulk protein and a relatively mild (up to ∼5%) improvement in bulk oil prediction. The unique combination of metric-MLR modeling approach (which is rare in the field of grain analysis) and sample treatments resulted in improved prediction models; using the physical attributes of individual grains is suggested as a novel measure for improving accuracy in prediction.
... Infrared (IR) spectroscopy is a modern analytical technique enabling molecular imaging of complex samples. This method is based on the absorption of infrared radiation by vibrational transitions in covalent bonds [31]. It allows for the acquisition of high-resolution images of the in-situ distribution of various molecules including carbohydrates, proteins, lipids, cholesterols, phospholipids, nucleic acids and small molecules. ...
Background
Dark pigmented lesions of the oral mucosa can represent a major diagnostic challenge. A biopsy is usually required to determine the nature of such intraoral discolorations. This study investigates the potential use of infrared spectroscopy for differential diagnosis of amalgam tattoos versus benign or malignant melanocytic neoplasms.
Materials and methods
For this retrospective study, formalin-fixed paraffin-embedded tissue (FFPE) specimens of dark pigmented lesions concerning the oral mucosa or the lip were investigated using mid infrared spectroscopy. The samples were chosen from patients who had undergone a mucosal biopsy at the University Hospital Innsbruck (Austria) between the years 2000 and 2017. Principal component analysis was used for data exploration. Evaluation was based on the superimposition of the recorded spectra and the corresponding histologic slides.
Results
In total, 22 FFPE specimens were analyzed. Clear differences were found between amalgam and non-amalgam samples. A general weakening of the penetrating infrared radiation allowed for unspecific discrimination between these two classes. An overall accuracy in predicting the correct class of 95.24% was achieved.
Conclusion
Infrared spectroscopy appears to be a suitable technique to differentiate between amalgam tattoos and melanocytic lesions in FFPE samples. It could potentially be applied in vivo, too, serving as a non-invasive diagnostic tool for intraoral dark pigmented lesions.
... Bayesian classifiers can be trained quickly for preliminary results, and were some of the first techniques used for classification of tissue biopsies imaged using FT-IR microscopy. 226,227 The major limitation for this type of classifier is often characterized as its assumption that all input features are independent. While this assumption is rarely true, these classifiers are routinely used for preliminary classification results owing to their speed and simplicity. ...
Vibrational spectroscopy and imaging promise molecular
information that can be rapidly acquired without the
need for specialized stains or dyes, thereby potentially
simplifying and speeding up necessary analyses for interventions
in many facets of modern day healthcare. The
salient characteristics of vibrational spectroscopy for
molecular analyses, using non-perturbative optical measurements,
and employing computational analysis of data,
are especially useful near the point of care as assessments
can be made with fewer reagents, under pressure of time
and accuracy while not requiring extensive specialized
human expertise. Significant technological development
has occurred and many seminal proof of concept studies
have been conducted to demonstrate the utility and
vast potential of spectroscopic methods. Accordingly,
a number of studies have focused on pushing the fundamental
performance limits of spectroscopic methods while
others have focused on specific problems where the use
of vibrational spectroscopy promises to change the standard
of care. Despite this impressive progress, however,
the application area is still maturing and rapidly evolving.
A vast array of potential applications continues to be
assessed while others need further technological developments.
In this review, we focus on recent developments
that demonstrate potential for point of care impact and
major trends that can lead, in turn, to improved spectroscopic
technology. We provide focused examples of ‘‘case
studies’’ and major trends in spectroscopic analyses ranging
from in vivo measurements to that of ex vivo bodily
fluids to extracted and processed tissues. In each case, the
uniting theme is that information to the clinician is enabled
closer to the patient, allowing for a shorter time between
identification of the need for analyses and availability of
information that guides care.
... The instrumentation and methods described here can also be used for other cases, but critical features of each disease will be different. Another possible outcome of this result is the association of subcellular changes with disease in both breast and other tissues that was not as definitively established using older technology (36). The sensitivity of both epithelial recognition and stromal engagement with cancer presents increased opportunities for clinical translation on one hand and a detailed view of the sample histology for research on the other. ...
Significance
Cancer alters both the morphological and the biochemical properties of multiple cell types in a tissue. Generally, the morphology of epithelial cells is practical for routine disease diagnoses. Here, infrared spectroscopic imaging biochemically characterizes breast cancer, both epithelial cells and the tumor-associated microenvironment. Unfortunately, conventional spectral analyses are slow. Hence, we designed and built a laser confocal microscope that demonstrates a high signal-to-noise ratio for confident diagnoses. The instrument cuts down imaging time from days to minutes, making the technology feasible for research and clinical translation. Finally, automated human breast cancer biopsy imaging is reported in ∼1 hour, paving the way for routine research into the total tumor (epithelial plus microenvironment) properties and rapid, label-free diagnoses.
... FTIR provides a tissue-level view of the sample without the use of dyes or other reagents that are known to degrade RNA 22,23 . Furthermore, this technique has been previously used to provide label-free histology of prostate tissue with over 98% accuracy in determining cell types [24][25][26] . While IR imaging is usually performed on specialized substrates, here we made a small modification to make it compatible with our silicon microchips. ...
Here, we present a technique that performs on-chip picoliter real-time reverse transcriptase loop mediated isothermal amplification (RT-LAMP) reactions on a histological tissue section without any analyte purification while preserving the native spatial location of the nucleic acid molecules. We demonstrate this method by amplifying TOP2A messenger RNA (mRNA) in a prostate cancer xenograft with 100 µm spatial resolution and by visualizing the variation in threshold time of amplification across the tissue. The on-chip reaction was validated by mRNA fluorescence in situ hybridization (mFISH) from cells in the tissue section. The entire process, from tissue loading on microchip to results from RT-LAMP can be carried out in less than 2 h. We anticipate that this technique, with its ease of use, fast turnaround, and quantitative molecular outputs, would become an invaluable tissue analysis tool for researchers and clinicians in the biomedical arena.
... This imaging method measures the composition of cells and tissue in terms of biochemical ingredients, based on the absorption of infrared radiation by vibrational transitions in covalent bonds. 19,23,58 Obtained spectra demonstrate a 'molecular ngerprint' of the investigated area of the sample by providing a high dimension of chemical information. Thus, different structures or mechanisms, which are directly linked to pathology, can be observed. ...
Infections caused by mucormycetes are life-threating infections, particularly for patients suffering from immunosuppression or uncontrolled diabetes mellitus. The early diagnosis of this devastating infectious disease is essential to target antifungal therapy and to improve patient’s outcome. However, the diagnosis of mucormycoses remains challenging, as clinical signs and symptoms are unspecific and comprehensively evaluated diagnostic tools missing. Therefore, we performed a retrospective case study on formalin-fixed, paraffin-embedded (FFPE) tissue of patients suffering from invasive mucormycosis, to evaluate the suitability of mid-infrared (MIR) microscopic imaging for the detection and identification of mucormycetes in human tissue sections by biochemical changes. 8 tissue samples of 8 patients with proven invasive mucormycosis (IM) were measured with MIR in 3 replicas. Inclusion criteria were: positive culture and/or positive molecular identification of the causative agent by real-time PCR and positive histological findings. Archived FFPE blocks of patients suffering from IM were cut and stained with Grocott. Mucormycete-positive tissue sections were chosen for MIR analysis after deparaffinization. MIR is a vibrational spectroscopic technique that uses infrared radiation to image proteins and small molecules by in-situ analysis in topographic maintained tissue sections. MIR detects and characterizes molecules formed by the interplay of host and microbial cells. We found that MIR is able to differentiate fungal elements from human tissue independent of the organ type studied. Fungal cells were detected and identified with MIR.
... In combination with multivariate analysis, it is possible to extract from large spectral data and correlate with biological information (characteristics, grade of tumors). Several previous reviews have already described in detail [7,73,74]. After collecting the data, it need to Pre-process which includes; I: Quality checks of the recorded data will be generated, including a) water vapor contribution, b) signal-to-noise-ratio c) absorbances in the range of the linear response of the detector. ...
Background:
Cancer is a major global health issue. It causes extensive individual suffering and gives a huge burden on the health care in society. Despite extensive research and different tools have been developed it still remains a challenge for early detection of this disease. FTIR imaging has been used to diagnose and differentiate the molecular differences between normal and diseased tissues.
Methods:
Fourier Transform Infrared Spectroscopy (FTIR) is able to measure biochemical changes in tissue, cell and biofluids based on the vibrational signature of their components. This technique enables to the distribution and structure of lipids, proteins, nucleic acids as well as other metabolites. These differences depended on the type and the grade of cancer.
Results:
We emphasize here, that the FTIR spectroscopy and imaging can be considered as a promising technique and will find its place on the detection of this dreadful disease because of high sensitivity, accuracy and inexpensive technique. Now the medical community started using and accepting this technique for early stage cancer detection. We discussed this technique and the several challenges in its application for the diagnosis of cancer in regards of sample preparations, data interpretation, and data analysis. The sensitivity of chemotherapy drugs on individual specific has also discussed.
Conclusion:
So far progressed has done with the FTIR imaging in understanding of cancer disease pathology. However, more research is needed in this field and it is necessary to understand the morphology and biology of the sample before using the spectroscopy and imaging because invaluable information to be figured out.
... These modern analytical techniques enable molecular imaging of complex samples and are based on the absorption of infrared radiation by vibrational transitions in covalent bonds [24]. The major advantage of these methods is the acquisition of unique images of the in situ distribution of proteins, lipids, carbohydrates, cholesterols, nucleic acids, phospholipids and small molecules with high spatial resolution whilst maintaining the topographic integrity of the tissue and avoiding time-consuming extraction, purification and separation steps [25]. ...
Due to the influence of many environmental processes, a precise determination of the post-mortem interval (PMI) of skeletal remains is known to be very complicated. Although methods for the investigation of the PMI exist, there still remains much room for improvement. In this study the applicability of infrared (IR) microscopic imaging techniques such as reflection-, ATR- and Raman- microscopic imaging for the estimation of the PMI of human skeletal remains was tested. PMI specific features were identified and visualized by overlaying IR imaging data with morphological tissue structures obtained using light microscopy to differentiate between forensic and archaeological bone samples. ATR and reflection spectra revealed that a more prominent peak at 1042 cm⁻¹ (an indicator for bone mineralization) was observable in archeological bone material when compared with forensic samples. Moreover, in the case of the archaeological bone material, a reduction in the levels of phospholipids, proteins, nucleic acid sugars, complex carbohydrates as well as amorphous or fully hydrated sugars was detectable at (reciprocal wavelengths/energies) between 3000 cm⁻¹ to 2800 cm⁻¹. Raman spectra illustrated a similar picture with less ν2PO4³⁻at 450 cm⁻¹ and ν4PO4³⁻ from 590 cm⁻¹ to 584 cm⁻¹, amide III at 1272 cm⁻¹ and protein CH2 deformation at 1446 cm⁻¹ in archeological bone material/samples/sources. A semi-quantitative determination of various distributions of biomolecules by chemi-maps of reflection- and ATR- methods revealed that there were less carbohydrates and complex carbohydrates as well as amorphous or fully hydrated sugars in archaeological samples compared with forensic bone samples. Raman- microscopic imaging data showed a reduction in B-type carbonate and protein α-helices after a PMI of 3 years. The calculated mineral content ratio and the organic to mineral ratio displayed that the mineral content ratio increases, while the organic to mineral ratio decreases with time. Cluster-analyses of data from Raman microscopic imaging reconstructed histo-anatomical features in comparison to the light microscopic image and finally, by application of principal component analyses (PCA), it was possible to see a clear distinction between forensic and archaeological bone samples. Hence, the spectral characterization of inorganic and organic compounds by the afore mentioned techniques, followed by analyses such as multivariate imaging analysis (MIAs) and principal component analyses (PCA), appear to be suitable for the post mortem interval (PMI) estimation of human skeletal remains.
... In biomedical studies involving disease screening, there naturally exists variation between individuals and possible confounding variables between samples [55][56][57] . Therefore screening programmes using spectroscopy must be specific enough to determine signatures attributable to a particular disease state in spite of 'noise' in the data. ...
Worldwide amphibian populations are declining due to habitat loss, disease and pollution. Vulnerability to environmental contaminants such as pesticides will be dependent on the species, the sensitivity of the ontogenic life stage and hence the timing of exposure and the exposure pathway. Herein we investigated the biochemical tissue ‘fingerprint’ in spawn and early-stage tadpoles of the Common frog, Rana temporaria, using attenuated total reflection-Fourier-transform infrared (ATR-FTIR) spectroscopy with the objective of observing differences in the biochemical constituents of the respective amphibian tissues due to varying water quality in urban and agricultural ponds. Our results demonstrate that levels of stress (marked by biochemical constituents such as glycogen that are involved in compensatory metabolic mechanisms) can be observed in tadpoles present in the pond most impacted by pollution (nutrients and pesticides), but large annual variability masked any inter-site differences in the frog spawn. ATR-FTIR spectroscopy is capable of detecting differences in tadpoles that are present in selected ponds with different levels of environmental perturbation and thus serves as a rapid and cost effective tool in assessing stress-related effects of pollution in a vulnerable class of organism.
... For pixel-based classification, the patients were split into a training group with 7 representatives of each disease subtype and a validation group with 3 of each disease subtype. The spectral data were converted to a set of 152 pre-defined metrics corresponding to spectral characteristics such as peak height ratios, normal areas, peak area ratios, and centers of gravity [13]. The training data were then exported for use with an algorithm in MATLAB to select metrics best suited to distinguishing between the disease states, and used to train a Naïve Bayesian classifier. ...
Fourier transform infrared (FT-IR) microscopy was used to image tissue samples from twenty patients diagnosed with thyroid carcinoma. The spectral data were then used to differentiate between follicular thyroid carcinoma and follicular variant of papillary thyroid carcinoma using principle component analysis and a Naïve Bayesian classifier operating on a set of computed spectral metrics. Classification of patients’ disease type was accomplished by using average spectra from a wide region containing follicular cells, colloid, and fibrosis; however, classification of disease state at the pixel level was only possible when the extracted spectra were limited to follicular epithelial cells in the samples, excluding the relatively uninformative areas of fibrosis. The results demonstrate the potential of FT-IR microscopy as a tool to assist in the difficult diagnosis of these subtypes of thyroid cancer, and also highlights the importance of selectively and separately analyzing spectral information from different features of a tissue of interest.
... For example, two vibrational modes are sufficient to visualize the structure in a two-phase polymer blend (13) and tens of frequencies may be needed to extract histopathologic information from tissue samples. New technological advances along with the realization to only record specific frequencies that may be needed (14) has led to the emergence of the discrete-frequency infrared (DFIR) imaging (15) approach. Here, we review current technology with a special emphasis on recent developments in discrete-frequency imaging in both IR and Raman. ...
Recent advances in instrumentation have enabled new forms of vibrational chemical imaging, including
discrete-frequency infrared (DFIR) microscopy and stimulated Raman scattering (SRS) microscopy. These technologies may represent a fundamental shift in how we approach spectroscopic imaging:
Rather than collecting full spectra that contain redundant information, measuring a few important spectral frequencies may enable significant gains in speed, throughput, signal-to-noise ratio,
and image quality. For IR microscopy, these advantages may be compounded by high-definition IR
microscopy. Here we discuss recent advances in IR and nonlinear Raman imaging through the lens of
“discrete-frequency” approaches and include several examples of applications and critical issues in
instrumentation that are likely to be dominant research themes in the near future.
... In order to quantify the nuclear and cellular morphology of epithelial and stromal cells and lumens (Fig. 2a), we first segment epithelium and stroma in tissue by adopting Fourier transform infrared (FT-IR) spectroscopy imaging due to its accuracy and robustness [44]. FT-IR has been extensively validated in classifying histologic cell types in tissue [49,60,61] and provides a color coded cell type image of tissue. Cell type segmentation in H&E images is challenging due to limited information, color variations, etc. Rigid-body image registration overlays the epithelium and stroma segmentation from FT-IR imaging with the corresponding H&E image by using outer shape and empty space (lumens) in tissues [45]. ...
Background
The current practice of histopathology review is limited in speed and accuracy. The current diagnostic paradigm does not fully describe the complex and complicated patterns of cancer. To address these needs, we develop an automated and objective system that facilitates a comprehensive and easy information management and decision-making. We also develop a tissue similarity measure scheme to broaden our understanding of tissue characteristics.
Results
The system includes a database of previously evaluated prostate tissue images, clinical information and a tissue retrieval process. In the system, a tissue is characterized by its morphology. The retrieval process seeks to find the closest matching cases with the tissue of interest. Moreover, we define 9 morphologic criteria by which a pathologist arrives at a histomorphologic diagnosis. Based on the 9 criteria, true tissue similarity is determined and serves as the gold standard of tissue retrieval. Here, we found a minimum of 4 and 3 matching cases, out of 5, for ~80 % and ~60 % of the queries when a match was defined as the tissue similarity score ≥5 and ≥6, respectively. We were also able to examine the relationship between tissues beyond the Gleason grading system due to the tissue similarity scoring system.
Conclusions
Providing the closest matching cases and their clinical information with pathologists will help to conduct consistent and reliable diagnoses. Thus, we expect the system to facilitate quality maintenance and quality improvement of cancer pathology.
Electronic supplementary material
The online version of this article (doi:10.1186/s12859-016-1086-6) contains supplementary material, which is available to authorized users.
... The intrinsic nature of the Fourier-transform and interferometry, however, necessitates that all wavelengths are acquired regardless of whether they are useful for final diagnostics or not. The discrete frequency infrared (DF-IR) spectroscopic microscopy technique is advantageous when the specific set of wavelengths that can discriminate the sample are already known 10,11 . Consequently, only the wavelengths useful for analysis are acquired thus resulting in a significant improvement in data efficiency. ...
Infrared (IR) spectroscopic imaging is an emerging modality for biological tissue analysis that has traditionally employed an interferometer for spectral discrimination. Recent technology developments have made discrete frequency sources, both lasers and filters, practical for imaging. The use of quantum cascade lasers in particular, presents new opportunities as well as challenges. Here we describe results from a novel point scanning confocal IR microscope and demonstrate the performance imaging several important spectral bands of lung tissue. Results show the possibility of discrete frequency (DF) absorbance measurements with RMS noise levels down to 0.34 mAU in 0.25 ms.
Cell cycle progression plays a vital role in regulating proliferation, metabolism, and apoptosis. Three-dimensional (3D) cell cultures have emerged as an important class of in vitro disease models, and incorporating the variation occurring from cell cycle progression in these systems is critical. Here, we report the use of Fourier transform infrared (FT-IR) spectroscopic imaging to identify subtle biochemical changes within cells, indicative of the G1/S and G2/M phases of the cell cycle. Following previous studies, we first synchronized samples from two-dimensional (2D) cell cultures, confirmed their states by flow cytometry and DNA quantification, and recorded spectra. We determined two critical wavenumbers (1059 and 1219 cm-1) as spectral indicators of the cell cycle for a set of isogenic breast cancer cell lines (MCF10AT series). These two simple spectral markers were then applied to distinguish cell cycle stages in a 3D cell culture model using four cell lines that represent the main stages of cancer progression from normal cells to metastatic disease. Temporal dependence of spectral biomarkers during acini maturation validated the hypothesis that the cells are more proliferative in the early stages of acini development; later stages of the culture showed stability in the overall composition but unique spatial differences in cells in the two phases. Altogether, this study presents a computational and quantitative approach for cell phase analysis in tissue-like 3D structures without any biomarker staining and provides a means to characterize the impact of the cell cycle on 3D biological systems and disease diagnostic studies using IR imaging.
Recent technological advances in IR imaging were the result of the coupling of a microscope with FTIR instrument. Acquiring spatially resolved chemical information has allowed significant advances in understanding cancer pathophysiology and discovering cancer biomarkers and anticancer drugs. Infrared spectroscopy coupled with microscopy has been reported to be a valuable nondestructive, label-free, minimal sample preparation, and highly sensitive analytical tool for biomedical research. Changes in biological tissue from normal to cancerous tissue are accompanied by changes in biomolecular composition and distribution. The identification and quantification of these specific biochemical changes could provide information that can be used for diagnosis and detection of cancerous tissues and tumors. This chapter reviews selected applications of FTIR microspectroscopy in the field of cancer research. Interests have been on new techniques for cancer grading, diagnosis of benign and malignant lesions in breast tissues, liver tumor detection, prostate cancer, and skin cancer.
Infrared spectral pathology has gained significant attention in the last few years, since it has been demonstrated to be able to readily identify cancerous tissue in biopsy samples. The Infrared technique, however, normally requires tissue sections to be mounted on infrared transparent slides. Unfortunately, these slides are both expensive and particularly frangible. In addition, mounting samples on specialist slides is an additional step in the sample preparation workflow, which ideally should be avoided. Applying infrared imaging directly to the H&E stained tissue on the glass slides that are normally used by pathologists, could help the infrared imaging technique be incorporated into current cancer diagnosis work flow and lower the total cost of detection. The disadvantage of using glass slides is that the spectral range available is restricted to just the high wavenumber region (2500-3600 cm⁻¹). In this work a study has been conducted on 120 breast tissues biopsy cores from different patients, to demonstrate that with the limited spectral information, breast cancer can be identified from the H&E glass slides. A four-class histological Adboost classification model has been constructed. Optimisation of the classification threshold was carried out to reduce the number of false negatives. Using a threshold of 0.1 the cancerous cores could be detected with an accuracy of 95.8%. This was incorporated into a simple traffic light system that could be used as a prescreening tool. This work, demonstrating the use of infrared spectral pathology on standard pathology samples slide, thus goes some way to overcome one of the barriers to successful translation of the infrared technique into the clinic.
Deep neural networks have emerged as a set of robust machine learning tools for computer vision. The suitability of convolutional and recurrent neural networks, along with their variants, is well documented for color image analysis. However, remote sensing and biomedical imaging often rely on hyperspectral images containing more than three channels for pixel-level characterization. Deep learning can facilitate image analysis in multi-channel images; however, network architecture and design choices must be tailored to the unique characteristics of this data. In this two-part series, we review convolution and recurrent neural networks as applied to hyperspectral imagery. Part I focuses on the algorithms and techniques, while Part II focuses on application-specific design choices and real-world remote sensing and biomedical test cases. These chapters also survey recent advances and future directions for deep learning with hyperspectral images.
Results of studies of the use of FTIR spectroscopy in cancer diagnosis are reviewed. Materials studied include biological fluids, tissues, and model systems for studying experimental neoplasia. The methods for working with each system are described. A detailed description is given for the use of IR spectroscopy for cancer diagnosis and in the monitoring of cancer therapy for the evaluation of drug effectiveness and determination of the disease state. Statistical methods for processing the IR spectral data are presented. The main limitations to the use of IR spectroscopy for the diagnosis of oncological diseases and the potential for its introduction into clinical practice are described.
Advancement of discrete frequency infrared (DFIR) spectroscopic microscopes in image quality and data throughput are critical to their use for analytical measurements. Here we report the development and characterization of a point scanning instrument with minimal aberrations and capable of diffraction-limited performance across all fingerprint region wavelengths over arbitrarily large samples. The performance of this system is compared to commercial state of the art Fourier transform infrared (FT-IR) imaging systems. We show that for large samples or smaller set of discrete frequencies, point scanning far exceeds (~10-100 fold) comparable data acquired with FT-IR instruments. Further we show improvements in image quality using refractive lenses that show significantly improved contrast across the spatial frequency bandwidth. Finally, we introduce the ability to image two tunable frequencies simultaneously using a single detector by means of demodulation to further speed up data acquisition and reduce the impact of scattering. Together, the advancements provide significantly better spectral quality and spatial fidelity than current state of the art imaging systems while promising to make spectral scanning even faster.
Current methods for cancer detection rely on tissue biopsy, chemical labeling/staining, and examination of the tissue by a pathologist. Though these methods continue to remain the gold standard, they are non-quantitative and susceptible to human error. Fourier transform infrared (FTIR) spectroscopic imaging has shown potential as a quantitative alternative to traditional histology. However, identification of histological components requires reliable classification based on molecular spectra, which are susceptible to artifacts introduced by noise and scattering. Several tissue types, particularly in heterogeneous tissue regions, tend to confound traditional classification methods. Convolutional neural networks (CNNs) are the current state-of-the-art in image classification, providing the ability to learn spatial characteristics of images. In this paper, we demonstrate that CNNs with architectures designed to process both spectral and spatial information can significantly improve classifier performance over per-pixel spectral classification. We report classification results after applying CNNs to data from tissue microarrays (TMAs) to identify six major cellular and acellular constituents of tissue, namely adipocytes, blood, collagen, epithelium, necrosis, and myofibroblasts. Experimental results show that the use of spatial information in addition to the spectral information brings significant improvements in the classifier performance and allows classification of cellular subtypes, such as adipocytes, that exhibit minimal chemical information but have distinct spatial characteristics. This work demonstrates the application and efficiency of deep learning algorithms in improving the diagnostic techniques in clinical and research activities related to cancer.
Tissue histology utilizing chemical and immunohistochemical labels plays an extremely important role in biomedicine and disease diagnosis. Recent research suggests that mid-infrared (IR) spectroscopic imaging may augment histology by providing quantitative molecular information. One of the major barriers to this approach is long acquisition time using Fourier-transform infrared (FTIR) spectroscopy. Recent advances in coherent sources, particularly quantum cascade lasers (QCLs), may mitigate this problem by allowing selective sampling of the absorbance spectrum. However, DFIR only provides a significant advantage when the number of spectral samples is minimized, requiring a priori knowledge of important spectral features. In this paper, we demonstrate the use of a GPU-based genetic algorithm (GA) using linear discriminant analysis (LDA) for DFIR feature selection. Our proposed method relies on pre-acquired broadband FTIR images for feature selection. Based on user-selected criteria for classification accuracy, our algorithm provides a minimal set of features that can be used with DFIR in a time-frame more practical for clinical diagnosis.
Optical chemical imaging seeks to non-destructively acquire spatially resolved chemical information without the use of labels or probes. The combination of vibrational mid-infrared (IR) spectroscopy and microscopy is especially attractive since the fundamental vibrational modes of samples are coincident with optical frequencies, thereby absorbing a large fraction of incident light and providing a strong signal. Modern IR spectroscopic imaging can be traced back nearly 25 years, with the coupling of an IR microscope to an array detector and an interferometer1,2. Instrumentation had been largely similar to this initial setup for almost two decades with several innovations such as rapid scan imaging3,4, time-resolved imaging⁵, faster detectors and linear array systems.⁶ These advances have provided several variants to speed up data acquisition and enable new capabilities compared to the basic configuration of a broadband globar source, interferometer and array detector-equipped microscope. Recent advances in hardware and design have dramatically changed both instrumentation and availability over the past few years. The availability of new components has led to a diversity in instrumentation, new understanding of image formation by rigorous theory has led to new designs and, consequently, novel applications have resulted in notable progress. A tremendous expansion in capability and exciting new possibilities for analytical measurements has now become apparent. Here we review these advances and organize the various developments in the framework of transforming the analytical performance of IR imaging in terms of spatial, temporal, and information content.
Breast cancer (BC) is a highly heterogeneous disease, both at the pathological and molecular level, and several chromatin-associated proteins play crucial roles in breast cancer initiation and progression. Here, we demonstrate the role of PSIP1 (PC4 and SF2 interacting protein)/p75 (LEDGF) in breast cancer progression. PSIP1/p75, previously identified as a chromatin-adaptor protein, is found to be upregulated in basal-like/triple negative breast cancer (TNBC) patient samples and cell lines. Immunohistochemistry in tissue arrays showed elevated levels of PSIP1 in metastatic invasive ductal carcinoma. Survival data analyses revealed that the levels of PSIP1 showed a negative association with TNBC patient survival. Depletion of PSIP1/p75 significantly reduced the tumorigenicity and metastatic properties of TNBC cell lines while its over-expression promoted tumorigenicity. Further, gene expression studies revealed that PSIP1 regulates the expression of genes controlling cell-cycle progression, cell migration, and invasion. Finally, by interacting with RNA polymerase II, PSIP1/p75 facilitates the association of RNA pol II to the promoter of cell cycle genes and thereby regulates their transcription. Our findings demonstrate an important role of PSIP1/p75 in TNBC tumorigenicity by promoting the expression of genes that control the cell cycle and tumor metastasis.
IR spectra of heme and different O2-content hemoglobin were studied by the quantum computation method at the molecule level. IR spectra of heme and different O2-content hemoglobin were quantificationally characterized from 0 to 100 THz. The IR spectra of oxy-heme and de-oxy-heme are obviously different at the frequency regions of 9.08-9.48, 38.38-39.78, 50.46-50.82, and 89.04-91.00 THz. At 24.72 THz, there exists the absorption peak for oxy-heme, whereas there is not the absorption peak for de-oxy-heme. Whether the heme contains Fe-O-O bond or not has the great influence on its IR spectra and vibration intensities of functional groups in the mid-infrared area. The IR adsorption peak shape changes hardly for different O2-content hemoglobin. However, there exist three frequency regions corresponding to the large change of IR adsorption intensities for containing-O2 hemoglobin in comparison with de-oxy-hemoglobin, which are 11.08-15.93, 44.70-50.22, and 88.00-96.68 THz regions, respectively. The most differential values with IR intensity of different O2-content hemoglobin all exceed 1.0???10(4)?L?mol(-1)?cm(-1). With the increase of oxygen content, the absorption peak appears in the high-frequency region for the containing-O2 hemoglobin in comparison with de-oxy-hemoglobin. The more the O2-content is, the greater the absorption peak is at the high-frequency region. The IR spectra of different O2-content hemoglobin are so obviously different in the mid-infrared region that it is very easy to distinguish the hemoglobin variant by means of IR spectra detector. IR spectra of hemoglobin from quantum computation can provide scientific basis and specific identification of hemoglobin variant resulting from different O2 contents in medical diagnosis.
Fourier transform infrared (FT-IR) spectroscopic imaging has shown great promise in becoming a powerful tool in cytology and histopathology. Applications for cancer diagnoses in solid tumors are especially attractive as samples are spatially complex and involve myriad molecular changes whereas there are many shortcomings in current clinical practice that can be addressed. Here we review the current state of the art in applying FT-IR imaging for analyzing solid tumors. We focus on instrumentation that is relatively new, emerging fundamental understanding gained by new theoretical advances, data analysis and selected, illustrative applications in cancer histopathology.
This chapter contains a short introduction to vibrational spectroscopy followed by an overview on its biological and biomedical applications. The spectroscopic techniques discussed in the book and their special advantages are briefly listed, i.e. Surface-Enhanced Raman Spectroscopy (SERS), Raman Optical Activity (ROA), Vibrational Circular Dichroism (VCD), Electronic Circular Dichroism (ECD) and matrix isolation.
The potential of vibrational spectroscopy is demonstrated by the current state of the art in secondary and primary plant components analysis performed in the tissue and from the single cells. Both Raman and IR spectroscopy are shown as powerful tools in medical diagnosis, cytology and histopathology. A brief overview on biomedical vibrational spectroscopy used to investigate lifestyle diseases is provided.
Fourier transform infrared (FT-IR) spectroscopic imaging has been widely tested as a tool for stainless digital histology of biomedical specimens, including for identification of infiltration and fibrosis in endomyocardial biopsy samples to assess transplant rejection. A major barrier in clinical translation has been the slow speed of imaging. To address this need, we tested and report here the viability of using high speed discrete frequency infrared (DFIR) imaging to obtain stain-free biochemical imaging in cardiovascular samples collected from patients. Images obtained by this method were classified with high accuracy by a Bayesian classification algorithm trained on FT-IR imaging data as well as on DFIR data. A single spectral feature correlated with instances of fibrosis, as identified by the pathologist, highlighting the advantage of the DFIR imaging approach for rapid detection. The speed of digital pathologic recognition was at least 16 times faster than the fastest FT-IR imaging instrument. These results indicate that a fast, on-site identification of fibrosis using IR imaging has potential for real time assistance during surgeries. Further, the work describes development and applications of supervised classifiers on DFIR imaging data, comparing classifiers developed on FT-IR and DFIR imaging modalities and identifying specific spectral features for accurate identification of fibrosis. This addresses a topic of much debate on the use of training data and cross-modality validity of IR measurements. Together, the work is a step towards addressing a clinical diagnostic need at acquisition time scales that make IR imaging technology practical for medical use.
Pathology is essential for research in disease and development, as well as for clinical decision making. For more than 100 years, pathology practice has involved analyzing images of stained, thin tissue sections by a trained human using an optical microscope. Technological advances are now driving major changes in this paradigm toward digital pathology (DP). The digital transformation of pathology goes beyond recording, archiving, and retrieving images, providing new computational tools to inform better decision making for precision medicine. First, we discuss some emerging innovations in both computational image analytics and imaging instrumentation in DP. Second, we discuss molecular contrast in pathology. Molecular DP has traditionally been an extension of pathology with molecularly specific dyes. Label-free, spectroscopic images are rapidly emerging as another important information source, and we describe the benefits and potential of this evolution. Third, we describe multimodal DP, which is enabled by computational algorithms and combines the best characteristics of structural and molecular pathology. Finally, we provide examples of application areas in telepathology, education, and precision medicine. We conclude by discussing challenges and emerging opportunities in this area.
This review focuses on chemometric techniques applied in MIR-biospectroscopy for cancer diagnosis and analysis during the last ten years of research. Experimental applications of chemometrics coupled with biospectroscopy are discussed throughout this work. The advantages and drawbacks of this association are also highlighted. Chemometric algorithms are evidenced as a powerful tool for cancer diagnosis, classification, and in different matrices. In fact, it is shown how chemometrics can be implemented along all different types of cancer analysis.
Biospectroscopic investigations have attracted attention of both the clinicians and basic sciences researchers in recent years. Scientists are discovering new areas for FTIR biospectroscopy applications in medicine. The aim of this study was to measure the possibility of FTIR-MSP application for the recognition and detection of fetus abnormalities after exposure of pregnant mouse to phenobarbital (PB) and levamisole (LEV) alone or in combination. PB is one of the most widely used antiepileptic drugs (AEDs), with sedative and hypnotic effects. When used by pregnant women, it is known to be a teratogenic agent. LEV is an antihelminthic drug with some applications in immune-deficiency as well as colon cancer therapy. Four groups of ten pregnant mice were selected for the experiments as follows: one control group received only standard diet, one group was injected with 120 mg/kg of BP, one group was injected with 10 mg/kg of LEV, and the last group was treated simultaneously with both BP and LEV at the above mentioned doses. Drugs administration was performed on gestation day 9 and fetuses were dissected on pregnancy day 15. Each dissected fetus was fixed, dehydrated and embedded in paraffin. Sections of liver (10 μm) were prepared from control and treated groups by microtome and deparaffinized with xylene. The spectra were taken by FTIR-MSP in the region of 4000–400 cm−1. All the spectra were normalized based on amide II band (1545 cm−1) after baseline correction of the entire spectrum, followed by classification using PCA, ANN and SVM. Both morphological and spectral changes were shown in the treated fetuses as compared to the fetuses in the control group. While cleft palate and C-R elongation were seen in PB injected fetuses, developmental retardation was mostly seen in the LEV injected group. Biospectroscopy revealed that both drugs mainly affected the cellular lipids and proteins, with LEV causing more changes in amide I and lipid regions than PB. Application of PCA, ANN and SVM methods were able to successfully classify these FTIR spectroscopic data and discriminate between control and treated groups of fetuses, making it a new potential tool for drugs teratogenic investigations.
Infrared (IR) spectroscopic imaging has been applied to study histology of cardiovascular tissue, primarily using Fourier transform IR (FTIR) Imaging. Here we describe results for histologic imaging of cardiac biopsies using a fast, discrete frequency IR (DFIR) imaging system. Histologic classification of tissue is understood in terms of the constituent frequencies and speeded up by careful optimization of the data acquired. Results are compared to FTIR imaging in terms of the signal to noise ratio and information content.
High-definition (HD) Fourier transform infrared (FT-IR) spectroscopic imaging is an emerging technique that not only enables chemistry-based visualization of tissue constituents, and label free extraction of biochemical information but its higher spatial detail makes it a potentially useful platform to conduct digital pathology. This methodology, along with fast and efficient data analysis, can enable both quantitative and automated pathology. Here we demonstrate a combination of HD FT-IR spectroscopic imaging of breast tissue microarrays (TMAs) with data analysis algorithms to perform histologic analysis. The samples comprise four tissue states, namely hyperplasia, dysplasia, cancerous and normal. We identify various cell types which would act as biomarkers for breast cancer detection and differentiate between them using statistical pattern recognition tools i.e. Random Forest (RF) and Bayesian algorithms. Feature optimization is integrally carried out for the RF algorithm, reducing computation time as well as redundant spectral features. We achieved an order of magnitude reduction in the number of features with comparable prediction accuracy to that of the original feature set. Together, the demonstration of histology and selection of features paves the way for future applications in more complex models and rapid data acquisition.
Optical spectroscopy and imaging have been used in medicine since medicine was first practised. However, the more sophisticated instrumental methods now under development have made little impact on clinical medicine. In this paper a brief overview of the development of optical diagnostics is presented, highlighted by some successful pre-clinical applications. The reasons for the slow penetration of optical diagnostics into clinical practice are discussed.
FT-IR microspectroscopic maps of unstained thin sections from human melanoma and colon carcinoma tissues were obtained on a conventional infrared microscope equipped with an automatic x, y stage. Mapped infrared data were analyzed by different image re-assembling techniques, namely functional group mapping ("chemical mapping") and, for the first time by cluster analysis, principal component analysis and artificial neural networks. The output values of the different classifiers were recombined with the original spatial information to construct IR-images whose color or gray tones were based on the spatial distribution of individual spectral patterns. While the functional group mapping technique could not reliably differentiate between the different tissue regions, the approach based on pattern recognition yielded images with a high contrast that confirmed standard histopathological techniques. The new technique turned out to be particularly helpful to improve discrimination between different types of tissue structures in general, and to increase image contrast between normal and cancerous regions of a given tissue sample.
Infrared spectra were obtained from exfoliated cervical cells from 156 females, of whom 136 were normal, 12 had cervical cancer, and 8 had dysplasia. The spectra of the normal women, essentially identical, differed from those obtained from patients with either cancer or dysplasia. In malignant samples we noted (i) significant changes in the intensity of the glycogen bands at 1025 cm-1 and 1047 cm-1, the bands at 1082 cm-1 and 1244 cm-1, the C--O stretching band at 1155 cm-1, and the band at 1303 cm-1, (ii) significant shifts of the peaks normally appearing at 1082 cm-1, 1155 cm-1, and 1244 cm-1, and (iii) an additional band at 970 cm-1. Further study of several of these bands, including the pressure dependence of their frequencies, revealed that in the malignant cervical tissue there were extensive changes in the degree of hydrogen bonding of phosphodiester groups of nucleic acids and C--OH groups of proteins, as well as changes in the degree of disorder of methylene chains of lipids. The IR spectra of samples with dysplasia demonstrated the same changes with cancer samples, except that the changes were of lesser magnitude and the phosphodiester peak normally appearing at 1082 cm-1 did not shift. These spectroscopic changes appear to progress in tandem with the morphological changes that lead normal cervical epithelium to cancer through the premalignant stage of dysplasia. The diagnostic potential of IR spectroscopy is discussed.
Using synchrotron radiation as an ultra-bright infrared source, we have been able to map the distributions of functional groups such as proteins, lipids, and nucleic acids inside a single living cell with a spatial resolution of a few microns. In particular, we have mapped the changes in the lipid and protein distributions in both the final stages of cell division and also during necrosis.
Various types of optical spectroscopy have been investigated as methods to effect a non-invasive, real-time in-situ assessment of tissue pathology. All of these methods have one basic principle in common: the optical spectrum of a tissue contains information about the biochemical composition and/or the structure of the tissue, and that information conveys diagnostic information. The biochemical information can be obtained by measuring absorption, fluorescence, or Raman scattering signals. Structural and morphological information may be obtained by techniques that assess the elastic-scattering properties of tissue. These basic approaches are useful for the detection of cancer as well as for other diagnostic applications such as hemoglobin saturation, intra-luminal detection of atherosclerosis, and simply the identification of different tissue types during procedures. Optical spectroscopic measurements can also be employed in the management of disease treatment. The site-specific pharmacokinetics of chemotherapy and photodynamic therapy agents can be used to customize dosage to the patient, and diagnostic spectroscopy can be used to monitor response to treatment. In recent years clinical studies have provided indications of potential efficacy, and some of these modalities are now entering a translational research stage, with an eye to approval and commercialization. A benefit of these methods is their inherent low cost and ease of implementation, generally mediated with small portable instruments, not requiring any specialized facilities, and eventually not requiring expert interpretation. This paper reviews briefly the most common methods of diagnostic optical spectroscopy, and reviews in greater depth recent clinical translational research invoking scattering spectroscopy as the enabling technology, which has been the experience of the authors.
Optical spectroscopy and imaging have been used in medicine since medicine was first practised. However, the more sophisticated instrumental methods now under development have made little impact on clinical medicine. In this paper a brief overview of the development of optical diagnostics is presented, highlighted by some successful pre-clinical applications. The reasons for the slow penetration of optical diagnostics into clinical practice are discussed. The past Spectroscopy and imaging are, and always have been, at the core of medicine and biology. The earliest, and still the most common, medical diagnostic instrument is actually a highly advanced spectroscopic imaging system. This spectroscopic imaging system is capable of interrogating skin and determining the extent of blunt traumatic injury and alterations in blood flow, metabolism and tissue hydration (oedema). The imaging system has a self-adjusting aperture to control light levels and a self-focussing lens. The light-sensing device is an intricate arrangement of four types of exquisitely sensitive photosensors in a two-dimensional array of hundreds of millions of pixels covering an impressive spectral range. Spectroscopic images are acquired and transmitted to a
Statistical pattern recognition is a term used to cover all stages of an investigation from problem formulation and data collection through to discrimination and classification, assessment of results and interpretation. This chapter introduces some of the basic concepts in classification and describes the key issues. It presents two complementary approaches to discrimination, namely a decision theory approach based on calculation of probability density functions and the use of Bayes theorem, and a discriminant function approach. Many different forms of discriminant function have been considered in the literature, varying in complexity from the linear discriminant function to multiparameter nonlinear functions such as the multilayer perceptron. Regression is an important part of statistical pattern recognition. Regression analysis is concerned with predicting the mean value of the response variable given measurements on the predictor variables and assumes a model of the form. Bayes' theorem; regression analysis; statistical process control
Acquisition of large data sets from human tissues by infrared (IR) microscopy is now routine. However, processing such large data sets, which may contain more than 10 000 spectra, provides an enormous challenge. Overcoming this challenge and developing nonsubjective methods for the analysis of IR microscopic results remain the major hurdle to developing clinically useful applications. A three-step pattern recognition strategy based upon linear discriminant analysis has been developed for use as a search engine for tissue characterization. The three-step strategy includes a genetic algorithm-guided data reduction step, a classification step based upon linear discriminant analysis, and a final step in which the discriminant coefficients are converted into a visually appealing, nonsubjective representation of the distribution of each class throughout the tissue section. The application of this search engine in the characterization of tumor-bearing skin is demonstrated.
Infrared microspectroscopy of biopsied canine lymph cells and tissue was performed to investigate the possibility of using IR spectra coupled with multivariate classification methods to classify the samples as normal, hyperplastic, or neoplastic (malignant). IR spectra were obtained in transmission mode through BaF2 windows and in reflection mode from samples prepared on gold-coated microscope slides. Cytology and histopathology samples were prepared by a variety of methods to identify the optimal methods of sample preparation. Cytospinning procedures that yielded a monolayer of cells on the BaF2 windows produced a limited set of IR transmission spectra. These transmission spectra were converted to absorbance and formed the basis for a classification rule that yielded 100% correct classification in a cross-validated context. Classifications of normal, hyperplastic, and neoplastic cell sample spectra were achieved by using both partial least-squares (PLS) and principal component regression (PCR) classification methods. Linear discriminant analysis applied to principal components obtained from the spectral data yielded a small number of misclassifications. PLS weight loading vectors yield valuable qualitative insight into the molecular changes that are responsible for the success of the infrared classification. These successful classification results show promise for assisting pathologists in the diagnosis of cell types and offer future potential for in vivo IR detection of some types of cancer.
For phase-separated multicomponent polymeric systems, characterization of the interface between the components is particularly challenging. We have observed an optical effect in the infrared that can be used to image the interface specifically. This method yields images of the interfaces based on the interfaces showing apparent absorption arising from changes in refractive index at frequencies far from the specific frequencies associated with the components of the mixture. This method has been applied to multicomponent samples of polymer-dispersed liquid crystals where the nature of the interface can be specifically altered by the application of an electric potential across the sample. Effects of this optical phenomenon on spectra from such multicomponent systems are discussed, and factors that complicate quantitative analysis of data from interfacial regions have been pointed out.
Infrared spectroscopy is a convenient tool for determining the molecular conformations of proteins, lipids, and nucleic acids, as well as for identifying unknown chemical compounds. Previously, we have shown that infrared spectroscopy may be used to characterize the conformation of an abnormal protein found in pathologic specimens of medullary carcinoma of the thyroid. The technique described in that paper is of limited general utility, however, because it requires that the abnormal protein deposit constitute the vast majority of the tissue specimen examined by infrared spectroscopy. In addition, it requires that the specimen be mounted on a relatively expensive CaF2 crystal. Since it is customary in diagnostic pathology to retain such mounted specimens in perpetuity, a new crystal would be required for each specimen, resulting in relatively high sample preparation costs. The recent availability of high-optical-quality infrared microscopes eliminates the requirement that the abnormal material of interest constitute most of the specimen to be examined. In this note we describe a method for preparing samples which drastically reduces the cost of preparing a permanent specimen while, at the same time, making it somewhat easier to acquire optimal spectra.
Model experiments were conducted in an effort to quantitatively assess the extent of stray light, resulting from diffraction, in an FT-IR microscope system. The effects of stray light were studied under conditions employing different aperturing modes, aperture sizes, and wavelengths of light. Results and consequences of the findings are discussed with respect to the spatial resolution and quantitative integrity of the data obtainable in mapping analyses of multilayer polymer laminates.
A new technique produces images worth a thousand spectra.
Two Fourier transform infrared (FT-IR) spectrometers with microspectroscopic capabilities-one equipped with a focal plane array (FPA) mercury-cadmium-telluride (MCT) detector (imaging) and the other a single-element MCT detector (mapping)-are compared, Two samples, a one-dimensional diffusion system and a two-dimensional phase-separated system, were studied with both analytical tools. Diffusion regions and concentration gradients were determined in the diffusion sample. Phase-specific chemical composition data were obtained in the phase-separated sample. The utility of each instrument for specific cases and their respective limitations are illustrated. It is shown that the FPA system has the ability to sample a large sample area with high spatial resolution in a short time without adverse diffraction effects. Because of this very rapid sampling many dynamic processes can he studied which are otherwise poorly monitored by the point-by-point mapping technique. The spatial resolution achievable, the quality of data obtained [signal-to-noise ratio (SNR)], and temporal resolution between mapped areas are intimately connected in the mapping system. However, the imaging systems have a fixed spatial resolution (dependent on the optics and detector) and the SNR considerations determine temporal resolution. However. it was also shown that by aperturing to a localized region, the single-element detector system is able to arrive at phase concentration information more rapidly, with less cumbersome processing and with higher SNRs than is possible with the current FPA technology. The chemical information obtained from both instruments is, within a small experimental error, identical. This work demonstrates what can be accomplished with each analytical FT-IR instrument, and serves as a comparative study to demonstrate the value of each analytical tool in specific situations.
Changes in the structural proteins and hydration during aging is responsible for altered skin morphologic and mechanical properties manifested as wrinkling, sagging, loss of elasticity, or apparent dryness. To gain insight into the age-related alterations in protein conformation and water structure, we obtained Raman spectra from the sun-protected buttock skin representing chronologic aging and the sun-exposed forearm skin representing combined effects of photoaging and chronologic aging. Ten aged individuals (five men, five women; age range 74–87) and 10 control young individuals (five men, five women; age range 22–29) entered the study. In the photoaged forearm skin the positions of protein-specific amide I, amide III, and CH stretching bands were shifted, suggesting increased protein folding. In contrast, major changes were seen only in the amide I peak in chronologically aged skin. The intensity of the 3250 cm–1 OH stretching band was increased in photoaged skin (but not in chronologically aged skin) indicating an increased water content. R() representation of the low-frequency region of Raman spectra was applied to determine water structure. In the young skin and chronologically aged skin water was mostly present in the bound form. In the photoaged skin, however, an increase in intensity at 180 cm–1 was noted, which reflects an increase in the not-protein bound water (tetrahedron water clusters). In conclusion, it seems that proteins in the photoaged skin are more compact and interact with water to limited degree. Impairment in protein hydration may add to the understanding of ultrastructural, mechanical, and biochemical changes in structural proteins in the aged skin.Keywords: intrinsic aging, solar aging, Raman spectroscopy, vibrational spectroscopy
We examined variations of in vivo near-infrared Fourier transform Raman spectra on healthy human skin. Diurnal and day-to-day variation, variation between persons, variation between different spots in the same body region and variation in spectra obtained on the same spot was studied in 13 volunteers. In 140 volunteers we examined variations in Raman spectra caused by skin pigmentation. In our spectral analysis we concentrated on the wavenumbers and relative intensities of the water band around 3250 cm−1, the amide I and amide III bands in the regions around 1645–1680 and 1230–1300 cm−1, respectively, and on the wavenumbers of the strong bands at 1450 and 2940 cm−1 that were used as reference bands. The results showed a diurnal variation and interpersonal variation in relative water intensity. In relative amide I intensity there was a variation between persons and between repeated measurements on the same spot. Between persons the variation might arise from different hydration of proteins, but it is difficult to explain why this variation is seen when repeated measurements are made on the same spot. Day-to-day variations were seen in the relative amide III intensity. Variations in amide III band wavenumbers were seen between persons, probably because of different collagen structure. Pigmentation did not influence wavenumbers or intensities considerably, but a regression analysis showed a correlation between background height and pigmentation. The conclusion of our study is that Raman spectra obtained on healthy human skin are reproducible and show mostly small variations. Copyright © 2002 John Wiley & Sons, Ltd.
The spatial resolution for infrared microspectroscopy is investigated to determine the practical limits imposed by diffraction or optical aberrations. Quantitative results are obtained using high brightness synchrotron radiation, which serves as a diffraction-limited infrared ''point source'' for the microscope. The measured resolving power is in good agreement with diffraction theory, including a 30% improvement for a confocal optical arrangement. The diffraction calculation also shows how the confocal setup leads to better image contrast. The full width at half maximum of the instrument's resolution pattern is approximately /2 for this arrangement. One achieves this diffraction limit when the instrument's apertures define a region having dimensions equal to the wavelength of interest. While commercial microspectrometers are well corrected for optical aberrations allowing diffraction-limited results, the standard substrates used for supporting specimens introduce chromatic aberrations. An analysis of this aberration is also presented, and correction methods described. © 2001 American Institute of Physics.
The authors previously have shown by gas chromatography-mass spectrometry that the hydroxyl radical (.OH) induces alterations in the DNA base structure of the female breast, which are premalignant markers of breast cancer. Fourier transform-infrared (FT-IR)-spectroscopy also has a high potential for revealing a broad array of structural changes in DNA that may provide important new insight into breast cancer etiology and prediction.
DNA from normal reduction mammoplasty tissue, invasive ductal carcinoma, and nearby microscopically normal tissue was analyzed by FT-IR spectroscopy. Statistical models based on DNA spectral properties were developed and compared with a statistical model previously used with base modifications.
Substantial differences were found in the spectral properties of DNA from women with normal and cancerous breast tissue, indicating an ability to discriminate cancerous tissue from noncancerous tissue with a sensitivity and specificity of 83%. Most importantly, the normal population was divided into subgroups in which a nonrandom progression was identified and a cancer-like DNA phenotype that was highly correlated (r > or = 0.90) with that of the patients with cancer was exhibited in 59% of the women. The spectral data, which also were highly correlated with the base-model data, were used to establish a model for predicting the probability of breast cancer. Consistent with the high cancer reoccurrence rate in the ipsilateral breast, 8 of 10 of the microscopically normal tissue specimens remaining after tumor excision were classified as cancerous using this model.
Progressive structural changes in the DNA of the normal female breast, leading to a premalignant cancer-like phenotype in a high proportion of women, are the basis for a new paradigm for understanding the etiology of breast cancer and predicting its occurrence at early stages of oncogenesis. The results also suggest therapeutic strategies for potentially reversing the extent of DNA damage, which may be useful in disease prevention and treatment.
Raman spectroscopy has the potential to provide real-time, in situ diagnosis of breast cancer during needle biopsy or surgery via an optical fiber probe. Understanding the chemical/morphological basis of the Raman spectrum of breast tissue is a necessary step in developing Raman spectroscopy as a tool for in situ breast cancer diagnosis. To understand the relationship between the Raman spectrum of a sample of breast tissue and its disease state, near-infrared Raman spectroscopic images of human breast tissue were acquired using a confocal microscope. These images were then compared with phase contrast and hematoxylin- and eosin-stained images to develop a chemical/morphological model of breast tissue Raman spectra. This model fits macroscopic tissue spectra with a linear combination of basis spectra derived from spectra of the cell cytoplasm, cell nucleus, fat, β-carotene, collagen, calcium hydroxyapatite, calcium oxalate dihydrate, cholesterol-like lipid deposits and water. Each basis spectrum represents data acquired from multiple patients and, when appropriate, from a variety of normal and diseased states. The model explains the spectral features of a range of normal and diseased breast tissue samples, including breast cancer. It can be used to relate the Raman spectrum of a breast tissue sample to diagnostic parameters used by pathologists. Copyright © 2002 John Wiley & Sons, Ltd.
A bestselling classic reference, now expanded and updated to cover the latest instrumentation, methods, and applications. The Second Edition of Fourier Transform Infrared Spectrometry brings this core reference up to date on the uses of FT-IR spectrometers today. The book starts with an in-depth description of the theory and current instrumentation of FT-IR spectrometry, with full chapters devoted to signal-to-noise ratio and photometric accuracy. Many diverse types of sampling techniques and data processing routines, most of which can be performed on even the less expensive instruments, are then described. Extensively updated, the Second Edition: Discusses improvements in optical components. Features a full chapter on FT Raman Spectrometry. Contains new chapters that focus on different ways of measuring spectra by FT-IR spectrometry, including fourteen chapters on such techniques as microspectroscopy, internal and external reflection, and emission and photoacoustic spectrometry. Includes a new chapter introducing the theory of vibrational spectrometry. Organizes material according to sampling techniques. Designed to help practitioners using FT-IR capitalize on the plethora of techniques for modern FT-IR spectrometry and plan their experimental procedures correctly, this is a practical, hands-on reference for chemists and analysts. It's also a great resource for students who need to understand the theory, instrumentation, and applications of FT-IR.
Both infrared (IR) and Raman spectroscopy are emerging as powerful probes of biomedically relevant properties of tissue and biological fluids. From tentative first steps, this field of endeavor is now beginning to mature as the central conceptual and technical issues come into focus. Using representative examples mainly from our own research, the aim of the present article is to provide the reader with a brief overview of progress to date.
We report a computational method to remove or reduce dispersion artifacts from infrared microspectral data collected in transflection (reflection/absorption) mode. This artifact occurs along the edges of tissue samples, in particular if the tissue does not adhere well to the substrate. The method proposed for the removal of the artifact is similar to the phase correction used in standard Fourier transform infrared spectroscopy.
Fourier transform infrared spectroscopy has been applied to the investigation of synovial fluids (SFs) aspirated from arthritic joints. Significant differences, related to differences in the composition of the fluid as a result of the disease processes, were found between spectra of SFs from joints affected by rheumatoid arthritis, osteoarthritis, spondyloarthropathies, and meniscal injuries. Linear discriminant analysis with leave-one-out cross validation was used to classify 239 SF film spectra obtained from 86 patients. Using a patient-based approach, in which the consensus of results obtained from three spectra of each fluid was taken as the diagnosis, multivariate analysis successfully classified spectra into four classes, in excellent agreement with clinical diagnosis (96.5% correct classification). These results demonstrate that when combined with a properly trained classifier, infrared spectra of SF films can be used as an aid in the diagnosis of arthritic disorders.
Fourier-transform infrared spectroscopy (FT-IR) was applied to the study of tissue sections of human colorectal cancer. Pairs of tissue samples from colorectal cancer and histologically normal mucosa 5-10 cm away from the tumor were obtained from 11 patients who underwent partial colectomy. All cancer specimens displayed abnormal spectra compared with the corresponding normal tissues. These changes involved the phosphate and C-O stretching bands, the CH stretch region, and the pressure dependence of the CH2 bending and C = O stretching modes. Our findings indicate that in colonic malignant tissue, there are changes in the degree of hydrogen-bonding of (i) oxygen atoms of the backbone of nucleic acids (increased); (ii) OH groups of serine, tyrosine, and threonine residues (any or all of them) of cell proteins (decreased); and (iii) the C = O groups of the acyl chains of membrane lipids (increased). In addition, they indicate changes in the structure of proteins and membrane lipids (as judged by the changes in their ratio of methyl to methylene groups) and in the packing and the conformational structure of the methylene chains of membrane lipids. The cell(s) of the malignant colon tissues responsible for these spectral abnormalities is unknown. Cultured colon adenocarcinoma cell lines displayed similarly abnormal FT-IR spectra. The diagnostic potential of the observed changes is discussed.
Diagnostic systems of several kinds are used to distinguish between two classes of events, essentially "signals" and "noise".
For them, analysis in terms of the "relative operating characteristic" of signal detection theory provides a precise and valid
measure of diagnostic accuracy. It is the only measure available that is uninfluenced by decision biases and prior probabilities,
and it places the performances of diverse systems on a common, easily interpreted scale. Representative values of this measure
are reported here for systems in medical imaging, materials testing, weather forecasting, information retrieval, polygraph
lie detection, and aptitude testing. Though the measure itself is sound, the values obtained from tests of diagnostic systems
often require qualification because the test data on which they are based are of unsure quality. A common set of problems
in testing is faced in all fields. How well these problems are handled, or can be handled in a given field, determines the
degree of confidence that can be placed in a measured value of accuracy. Some fields fare much better than others.
Infrared spectra of human central nervous system tissue and human breast carcinoma are presented. The spectra are discussed in terms of the composition of the tissues. It is shown that differences between spectra of white and grey matter can be rationalised on the basis of differences in lipid content. Spectra of the choroid plexus and arachnoid villus of the meninges show a series of absorptions not observed in other CNS tissue. These absorptions are discussed in terms of the connective tissue content of the samples. We demonstrate that the presence of collagen results in the appearance of a series of characteristic absorptions which may be mis-assigned as DNA phosphate absorptions. The implications of the presence of collagen in tissues for the diagnosis of disease states by IR spectroscopic methods, with particular reference to cancer, is discussed.
This study evaluates a variety of techniques and sampling conditions for Raman spectroscopic investigations of human stratum corneum. Using a Fourier-transform Raman system and samples of stratum corneum in vitro, we demonstrated minimal inter- and intracadaver variations in molecular vibrations. We have also shown Raman spectroscopy to be relatively insensitive to the hydration state of human stratum corneum, indicating that the technique should be valuable for monitoring the transdermal delivery of drugs from aqueous solutions. The stability of human stratum corneum to near-infrared laser excitation was verified by spectral collection for approximately 1 hr. We have also compared FT-Raman spectra from human stratum corneum in vitro and in vivo. Of the different types of Raman instruments used in this study (visible-light excitation microprobe, visible-light excitation macroscopic sampling, and Fourier-transform Raman), the FT-Raman system provided good-quality spectra with high sample throughput, but systems using visible-light excitation should provide unique information for use in specialist applications.
A powerful new mid-infrared spectroscopic chemical imaging technique combining step-scan Fourier transform Michelson interferometry with indium antimonide focal-plane array (FPA) image detection is described. The coupling of an infrared focal-plane array detector to an interferometer provides an instrumental multiplex/multichannel advantage. Specifically, the multiple detector elements enable spectra at all pixels to be collected simultaneously, while the interferometer portion of the system allows all the spectral frequencies to be measured concurrently. With this method of mid-infrared spectroscopic imaging, the fidelity of the generated spectral images is limited only by the number of pixels on the FPA detector, and only several seconds of starting time is required for spectral image acquisition. This novel, high-definition technique represents the future of infrared chemical imaging analysis, a new discipline within the chemical and material sciences, which combines the capability of spectroscopy for molecular analysis with the power of visualization. In particular, chemical imaging is broadly applicable for noninvasive, molecular characterization of heterogeneous materials, since all solid-state materials exhibit chemical nonuniformity that exists either by design or by development during the course of material preparation or fabrication. Imaging, employing Raman and infrared spectroscopy, allows the precise characterization of the chemical composition, domain structure, and chemical architecture of a variety of substances. This information is often crucial to a wide range of activities, extending from the fabrication of new materials to a basic understanding of biological samples. In this study, step-scan imaging principles, instrument design details, and infrared chemical imaging results are presented. Since the prospect of performing high-resolution and high-definition mid-infrared chemical imaging very rapidly has been achieved with the step-scan approach, the implications for the chemical analysis of materials are many and varied.
Many genes and signalling pathways controlling cell proliferation, death and differentiation, as well as genomic integrity, are involved in cancer development. New techniques, such as serial analysis of gene expression and cDNA microarrays, have enabled measurement of the expression of thousands of genes in a single experiment, revealing many new, potentially important cancer genes. These genome screening tools can comprehensively survey one tumor at a time; however, analysis of hundreds of specimens from patients in different stages of disease is needed to establish the diagnostic, prognostic and therapeutic importance of each of the emerging cancer gene candidates. Here we have developed an array-based high-throughput technique that facilitates gene expression and copy number surveys of very large numbers of tumors. As many as 1000 cylindrical tissue biopsies from individual tumors can be distributed in a single tumor tissue microarray. Sections of the microarray provide targets for parallel in situ detection of DNA, RNA and protein targets in each specimen on the array, and consecutive sections allow the rapid analysis of hundreds of molecular markers in the same set of specimens. Our detection of six gene amplifications as well as p53 and estrogen receptor expression in breast cancer demonstrates the power of this technique for defining new subgroups of tumors.
Noninvasive techniques that provide detailed information about molecular composition, structure, and interactions are crucial to further our understanding of the relation between skin disease and biochemical changes in the skin, as well as for the development of penetration enhancers for transdermal drug administration. In this study we present in vitro and in vivo Raman spectra of human skin. Using a Raman microspectrometer, in vitro spectra were obtained of thin cross sections of human skin. They provided insight into the molecular composition of different skin layers. Evidence was found for the existence of a large variation in lipid content of the stratum corneum. A simple experimental setup for in vivo confocal Raman microspectroscopy of the skin was developed. In vivo Raman spectra of the stratum corneum were obtained at different positions of the arm and hand of three volunteers. They provided evidence for differences in the concentration of natural moisturizing factor at these positions.
Infrared spectra of myeloid leukemia (ML-1) cells are reported for cells derived from an asynchronous, exponentially growing culture, as well as for cells that were fractionated according to their stage within the cell division cycle. The observed results suggest that the cells' DNA is detectable by infrared spectroscopy mainly when the cell is in the S phase, during the replication of DNA. In the G1 and G2 phases, the DNA is so tightly packed in the nucleus that it appears opaque to infrared radiation. Consequently, the nucleic acid spectral contributions in the G1 and G2 phases would be mostly that of cytoplasmic RNA. These results suggest that infrared spectral changes observed earlier between normal and abnormal cells may have been due to different distributions of cells within the stages of the cell division cycle.
To determine whether diagnostic information may be recovered from the infrared spectra of exfoliated cell specimens by using a novel spectral feature extraction method, in conjunction with linear and quadratic discriminant analysis, for spectral classification.
Over 800 infrared spectra were included in the study, with corresponding clinical diagnoses based upon cytology and, when available, histology reports. Three sets of classification trials were carried out with the aim of distinguishing the spectra corresponding to normal specimens from CIN 1, 2 and 3. For each of these three cases, the procedure was to: (1) develop a set of provisional classification models using only a "training" subset of the spectra, and (2) test each provisional model by its ability to correctly predict the diagnoses on the basis of the remaining spectra.
For optimal classification trials, training set classification accuracies were 68% for normal/CIN 1, 73% for normal/CIN 2 and 81% for normal/CIN 3; for the corresponding test sets the classification accuracies were 60%, 60% and 67%, respectively.
The infrared spectra of exfoliated cervical cells carry information regarding the presence or absence of dysplasia, and that information is recoverable--albeit imperfectly at this stage--from the spectra of "real life" cell preparations.
With the introduction of focal plane array detectors, FTIR microspectroscopy has significantly improved. Jack Koenig of Case Western Reserve University, Shi-Qing Wang with the University of Akron, and Rohit Bhargava from the National Institutes of Health describe the method, its instrumentation, and how it is being applied.
The signal-to-noise ratio (SNR) of spectral data obtained from a microimaging Fourier transform infrared (FT-IR) spectrometer assembly, employing a step-scan interferometer and focal plane array detector, is analyzed. Based on the methodology of data collection, a theoretical description for the performance characteristics is proposed and quantitative effects of the acquisition parameters on the SNR are explained theoretically and compared to experiment. To obtain the best strategy for achieving either the highest SNR in a given time interval or for attaining a given SNR in the shortest time period, the concept of characteristic plots is introduced. The theoretical analysis is extended to FT-IR microimaging employing continuous scan interferometers in which the advantages of fast image collection are enumerated, while SNR limitations arising from mirror positioning errors are discussed. A step-scan method is suggested for faster data collection in which an optimal detector response and SNR benefits are retained. Theoretically obtained SNRs based upon the expressions proposed in this paper predict experimentally determined values quite well and can be used to obtain an understanding of the required developments for improved performance. Finally, SNRs for both microimaging systems and conventional microspectroscopic instrumentation are compared.
High-density tissue microarrays (TMA) are useful for profiling protein expression in a large number of samples but their use for clinical biomarker studies may be limited in heterogeneous tumors like prostate cancer. In this study, the optimization and validation of a tumor sampling strategy for a prostate cancer outcomes TMA is performed. Prostate cancer proliferation determined by Ki-67 immunohistochemistry was tested. Ten replicate measurements of proliferation using digital image analysis (CAS200, Bacus Labs, Lombard, IL, USA) were made on 10 regions of prostate cancer from a standard glass slide. Five matching tissue microarray sample cores (0.6 mm diameter) were sampled from each of the 10 regions in the parallel study. A bootstrap resampling analysis was used to statistically simulate all possible permutations of TMA sample number per region or sample. Statistical analysis compared TMA samples with Ki-67 expression in standard pathology immunohistochemistry slides. The optimal sampling for TMA cores was reached at 3 as fewer TMA samples significantly increased Ki-67 variability and a larger number did not significantly improve accuracy. To validate these results, a prostate cancer outcomes tissue microarray containing 10 replicate tumor samples from 88 cases was constructed. Similar to the initial study, 1 to 10 randomly selected cores were used to evaluate the Ki-67 expression for each case, computing the 90th percentile of the expression from all samples used in each model. Using this value, a Cox proportional hazards analysis was performed to determine predictors of time until prostate-specific antigen (PSA) recurrence after radical prostatectomy for clinically localized prostate cancer. Examination of multiple models demonstrated that 4 cores was optimal. Using a model with 4 cores, a Cox regression model demonstrated that Ki-67 expression, preoperative PSA, and surgical margin status predicted time to PSA recurrence with hazard ratios of 1.49 (95% confidence interval [CI] 1.01-2.20, p = 0.047), 2.36 (95% CI 1.15-4.85, p = 0.020), and 9.04 (95% CI 2.42-33.81, p = 0.001), respectively. Models with 3 cores to determine Ki-67 expression were also found to predict outcome. In summary, 3 cores were required to optimally represent Ki-67 expression with respect to the standard tumor slide. Three to 4 cores gave the optimal predictive value in a prostate cancer outcomes array. Sampling strategies with fewer than 3 cores may not accurately represent tumor protein expression. Conversely, more than 4 cores will not add significant information. This prostate cancer outcomes array should be useful in evaluating other putative prostate cancer biomarkers.
Fourier transform IR (FTIR) microspectroscopy at a spatial resolution of 18 microm was used to study skin fibroblasts and giant sarcoma cells. Both cell lines were derived from the same patient; they were metabolically active and in the exponentially growing phase. The IR spectra were acquired for the nuclei and cytosol of untreated cells, cells washed with ethanol, and cells treated with RNase or DNase. A comparison of the spectra of the two cell lines yielded only insignificant spectral differences, indicating that IR spectroscopy monitors the overall cell activity rather than specific signs of cancer.
The recent development of tissue microarrays-composed of hundreds of tissue sections from different tumors arrayed on a single glass slide-facilitates rapid evaluation of large-scale outcome studies. Realization of this potential depends on the ability to rapidly and precisely quantify the protein expression within each tissue spot. We have developed a set of algorithms that allow the rapid, automated, continuous and quantitative analysis of tissue microarrays, including the separation of tumor from stromal elements and the sub-cellular localization of signals. Validation studies using estrogen receptor in breast carcinoma show that automated analysis matches or exceeds the results of conventional pathologist-based scoring. Automated analysis and sub-cellular localization of beta-catenin in colon cancer identifies two novel, prognostically significant tumor subsets, not detected by traditional pathologist-based scoring. Development of automated analysis technology empowers tissue microarrays for use in discovery-type experiments (more typical of cDNA microarrays), with the added advantage of inclusion of long-term demographic and patient outcome information.
Immunohistochemistry (IHC) of mismatch repair (MMR) proteins in colorectal tumors together with microsatellite analysis (MSI) can be helpful in identifying families eligible for mutation analysis. The aims were to determine sensitivity of IHC for MLH1, MSH2, and MSH6 and MSI analysis in tumors from known MMR gene mutation carriers; and to evaluate the use of tissue microarrays for IHC (IHC-TMA) of colon tumors in its ability to identify potential carriers of MMR gene mutations, and compare it with IHC on whole slides. IHC on whole slides was performed in colorectal tumors from 45 carriers of a germline mutation in one of the MMR genes. The TMA cohort consisted of 129 colon tumors from (suspected) hereditary nonpolyposis colorectal cancer (HNPCC) patients. Whole slide IHC analysis had a sensitivity of 89% in detecting MMR deficiency in carriers of a pathogenic MMR mutation. Sensitivity by MSI analysis was 93%. IHC can also be used to predict which gene is expected to harbor the mutation: for MLH1, MSH2, and MSH6, IHC on whole slides would have correctly predicted the mutation in 48%, 92%, and 75% of the cases, respectively. We propose a scheme for the diagnostic approach of families with (suspected) HNPCC. Comparison of the IHC results based on whole slides versus TMA, showed a concordance of 85%, 95%, and 75% for MLH, MSH2, and MSH6, respectively. This study therefore shows that IHC-TMA can be reliably used to simultaneously screen a large number of tumors from (suspected) HNPCC patients, at first in a research setting.
Instrumentation used in infrared microspectroscopy (IR-MSP) permits the acquisition of spectra from samples as small as 100 pg (10(-10) g), and as small as 1 pg for Raman microspectroscopy (RA-MSP). This, in turn, allows the acquisition of spectral data from objects as small as fractions of human cells, and of small regions of microtome tissue sections. Since vibrational spectroscopy is exquisitely sensitive to the biochemical composition of the sample, and variations therein, it is possible to monitor metabolic processes in tissue and cells, and to construct spectral maps based on thousands of IR spectra collected from pixels of tissue. These images, in turn, reveal information on tissue structure, distribution of cellular components, metabolic activity and state of health of cells and tissue.
In this paper, three different clustering algorithms were applied to assemble infrared (IR) spectral maps from IR microspectra of tissues. Using spectra from a colorectal adenocarcinoma section, we show how IR images can be assembled by agglomerative hierarchical (AH) clustering (Ward's technique), fuzzy C-means (FCM) clustering, and k-means (KM) clustering. We discuss practical problems of IR imaging on tissues such as the influence of spectral quality and data pretreatment on image quality. Furthermore, the applicability of cluster algorithms to the spatially resolved microspectroscopic data and the degree of correlation between distinct cluster images and histopathology are compared. The use of any of the clustering algorithms dramatically increased the information content of the IR images, as compared to univariate methods of IR imaging (functional group mapping). Among the cluster imaging methods, AH clustering (Ward's algorithm) proved to be the best method in terms of tissue structure differentiation.
The process of histopathology, comprising tissue staining and morphological pattern recognition, has remained largely unchanged for over 140 years. Although it is integral to clinical and research activities, histopathologic recognition remains a time-consuming, subjective process to which only limited statistical confidence can be assigned because of inherent operator variability. Although immunohistochemical approaches allow limited molecular detection, significant challenges remain in using them for quantitative, automated pathology. Vibrational spectroscopic approaches, by contrast, directly provide nonperturbing molecular descriptors, but a practical spectroscopic protocol for histopathology is lacking. Here we couple high-throughput Fourier transform infrared (FTIR) spectroscopic imaging of tissue microarrays with statistical pattern recognition of spectra indicative of endogenous molecular composition and demonstrate histopathologic characterization of prostatic tissue. This automated histologic segmentation is applied to routine archival tissue samples, incorporates well-defined tests of statistical significance and eliminates any requirement for dyes or molecular probes. Finally, we differentiate benign from malignant prostatic epithelium by spectroscopic analyses.
The recent development of Fourier transform infrared (FTIR) spectroscopic imaging has enhanced our capability to examine, on a microscopic scale, the spatial distribution of vibrational spectroscopic signatures of materials spanning the physical and biomedical disciplines. Recent activity in this emerging area has concentrated on instrumentation development, theoretical analyses to provide guidelines for imaging practice, novel data processing algorithms, and the introduction of the technique to new fields. To illustrate the impact and promise of this spectroscopic imaging methodology, we present fundamental principles of the technique in the context of FTIR spectroscopy and review new applications in various venues ranging from the physical chemistry of macromolecular systems to the detection of human disease.
Tissue microarrays are a high-throughput method for the investigation of biomarkers in multiple tissue specimens at once. This technique allows for the analysis of up to 500 tissue samples in a single experiment using immunohistochemistry and in situ hybridization. Recently, cell lines and xenografts have been reduced to a tissue microarray format and are being applied to preclinical drug development. In clinical research, tissue microarrays are applied at multiple levels: comprehensive analysis of samples in the context of a clinical trial or across a population. Tissue microarrays play a central role in translational research, facilitating the discovery of molecules that have potential roles in the diagnosis, prognosis and prediction of response to therapy.
An alternative approach has been suggested, in which reflective glass slides are employed ( Recently, spectral corrections have been proposed that allow for a better understanding of the distortion of spectra measured using these types of reflective slides
- Tj Leary
- Engler
- Tj Leary
- Engler
- I W Levin
- R Bhargava
I.W. Levin, R. Bhargava, Ann. Rev. Phys. Chem. 56 (2005) 429–474.
Various effects contribute to the spatial fidelity loss in mapping microscopes as described in Sommer AJ
Various effects contribute to the spatial fidelity loss in mapping
microscopes as described in Sommer AJ, Katon JE Appl. Spectrosc. 45:
1633–1640 (1991).
- A C Williams
- B W Barry
- H G M Edwards
A.C. Williams, B.W. Barry, H.G.M. Edwards, D.W. Farwell, Pharm. Res.
10 (1993) 1642.
- P J Caspers
- G W Lucassen
- R Wolthius
- H A Bruining
P.J. Caspers, G.W. Lucassen, R. Wolthius, H.A. Bruining, G.J. Puppels,
Biospectroscopy 4 (1998) S31.