Multi-class classification algorithm for optical diagnosis of oral cancer

We report development of a direct multi-class spectroscopic diagnostic algorithm for discrimination of high-grade cancerous tissue sites from low-grade as well as precancerous and normal squamous tissue sites of human oral cavity. The algorithm was developed making use of the recently formulated theory of total principal component regression (TPCR). The in vivo autofluorescence spectral data acquired from patients screened for neoplasm of oral cavity at the Government Cancer Hospital, Indore, was used to train and validate the algorithm. The diagnostic algorithm based on TPCR was found to provide satisfactory performance in classifying the tissue sites in four different classes - high-grade squamous cell carcinoma, low-grade squamous cell carcinoma, leukoplakia, and normal squamous tissue. The classification accuracy for these four classes was observed to be approximately 94%, 100%, 100% and 91% for the training data set (based on leave-one-out cross-validation), and was approximately 90%, 90%, 85% and 88%, respectively for the corresponding classes for the independent validation data set.     

Endoscopic light-induced autofluorescence spectroscopy for the diagnosis of colorectal cancer and adenoma

To evaluate the new, bio-optical method of light-induced autofluorescence spectroscopy for the endoscopic in-vivo diagnosis of (pre)-cancerous lesions of the colorectum, 311 endogenous fluorescence spectra were obtained from normal, adenomatous and cancerous colorectal tissue in 11 patients with cancer, six patients with familial adenomatous polyposis, and six patients with multiple adenomatous polyps. A light source delivered either white or violet-blue light for excitation of tissue autofluorescence via a flexible endoscope. Endogenous fluorescence spectra emitted by the tissue were picked up with a fiberoptic probe and analysed with a spectrograph. Biopsies were taken for definitive classification of the spectra. Rectal cancer (n=11) as well as adenomas with severe dysplasia (n=19) showed specific differences between the emitted fluorescence spectra as compared with normal mucosa and hyperplastic polyps. Having applied a mathematical algorithm to the spectra, a sensitivity of 96% and a specificity of 93% were obtained for the diagnosis of rectal cancer. The equivalent values for the diagnosis of dysplastic ademomas were 98 and 89%, respectively. Light-induced autofluorescence spectroscopy is a new and promising bio-optical procedure for the endoscopic in-vivo diagnosis of colorectal cancer and dysplasia.     

Optimal excitation wavelengths for in vivo detection of oral neoplasia using fluorescence spectroscopy

There is no satisfactory mechanism to detect premalignant lesions in the upper aero-digestive tract. Fluorescence spectroscopy has potential to bridge the gap between clinical examination and invasive biopsy; however, optimal excitation wavelengths have not yet been determined. The goals of this study were to determine optimal excitation-emission wavelength combinations to discriminate normal and precancerous/cancerous tissue, and estimate the performance of algorithms based on fluorescence. Fluorescence excitation-emission matrices (EEM) were measured in vivo from 62 sites in nine normal volunteers and 11 patients with a known or suspected premalignant or malignant oral cavity lesion. Using these data as a training set, algorithms were developed based on combinations of emission spectra at various excitation wavelengths to determine which excitation wavelengths contained the most diagnostic information. A second validation set of fluorescence EEM was measured in vivo from 281 sites in 56 normal volunteers and three patients with a known or suspected premalignant or malignant oral cavity lesion. Algorithms developed in the training set were applied without change to data from the validation set to obtain an unbiased estimate of algorithm performance. Optimal excitation wavelengths for detection of oral neoplasia were 350, 380 and 400 nm. Using only a single emission wavelength of 472 nm, and 350 and 400 nm excitation, algorithm performance in the training set was 90% sensitivity and 88% specificity and in the validation set was 100% sensitivity, 98% specificity. These results suggest that fluorescence spectroscopy can provide a simple, objective tool to improve in vivo identification of oral cavity neoplasia.     

Raman spectroscopy in combination with background near-infrared autofluorescence enhances the in vivo assessment of malignant tissues

The diagnostic ability of optical spectroscopy techniques, including near-infrared (NIR) Raman spectroscopy, NIR autofluorescence spectroscopy and the composite Raman and NIR autofluorescence spectroscopy, for in vivo detection of malignant tumors was evaluated in this study. A murine tumor model, in which BALB/c mice were implanted with Meth-A fibrosarcoma cells into the subcutaneous region of the lower back, was used for this purpose. A rapid-acquisition dispersive-type NIR Raman system was employed for tissue Raman and NIR autofluorescence spectroscopic measurements at 785-nm laser excitation. High-quality in vivo NIR Raman spectra associated with an autofluorescence background from mouse skin and tumor tissue were acquired in 5 s. Multivariate statistical techniques, including principal component analysis (PCA) and linear discriminant analysis (LDA), were used to develop diagnostic algorithms for differentiating tumors from normal tissue based on their spectral features. Spectral classification of tumor tissue was tested using a leave-one-out, cross-validation method, and the receiver operating characteristic (ROC) curves were used to further evaluate the performance of diagnostic algorithms derived. Thirty-two in vivo Raman, NIR fluorescence and composite Raman and NIR fluorescence spectra were analyzed (16 normal, 16 tumors). Classification results obtained from cross-validation of the LDA model based on the three spectral data sets showed diagnostic sensitivities of 81.3%, 93.8% and 93.8%; specificities of 100%, 87.5% and 100%; and overall diagnostic accuracies of 90.6%, 90.6% and 96.9% respectively, for tumor identification. ROC curves showed that the most effective diagnostic algorithms were from the composite Raman and NIR autofluorescence techniques.     

Discrimination of Osteonecrosis and Normal Tissues by Near-Infrared Spectroscopy and Successive Projections Algorithm-Linear Discriminant Analysis

Osteonecrosis of femoral head (ONFH) is a disease characterized by an impaired blood flow in the bone. The pathogenesis is still unknown, which makes an exact diagnosis troublesome and heavily dependent on experience. Exploring the information of molecular level by modern spectroscopy may help to discover the underlying pathogenesis and find its diagnostic application in clinical medicine. The study focuses on the combination of near-infrared (NIR) spectroscopy and classification models for discriminating ONFH and normal tissues. A total of 128 surgical specimens was prepared and NIR spectra were recorded by an integrating sphere. The experiment data set was divided into three subsets, i.e., the training set, validation set, and test set. Successive projection algorithm-linear discriminant analysis (SPA-LDA) was used to compress variables and build the diagnostic model. Partial least square-discriminant analysis (PLS-DA) was used as the reference. Principal component analysis (PCA) was used for exploratory analysis. The results showed that compared to PLS-DA, SPA-LDA provided a more parsimonious model using only seven variables and achieved better performance, i.e., sensitivity of 90.5 and 85%, and specificity of 100 and 95.5% for the validation and test sets, respectively. It indicated that NIR spectroscopy combined with SPA-LDA algorithm was a feasible aid tool for discriminating ONFH from normal tissue.     

High-wavenumber FT-Raman spectroscopy for in vivo and ex vivo measurements of breast cancer

The identification of normal and cancer breast tissue of rats was investigated using high-frequency (HF) FT-Raman spectroscopy with a near-infrared excitation source on in vivo and ex vivo measurements. Significant differences in the Raman intensities of prominent Raman bands of lipids and proteins structures (2,800?C3,100?cm^?1) as well as in the broad band of water (3,100?C3,550?cm^?1) were observed in mean normal and cancer tissue spectra. The multivariate statistical analysis methods of principal components analysis (PCA) and linear discriminant analysis (LDA) were performed on all high-frequency Raman spectra of normal and cancer tissues. LDA results with the leave-one-out cross-validation option yielded a discrimination accuracy of 77.2, 83.3, and 100% for in vivo transcutaneous, in vivo skin-removed, and ex vivo biopsy HF Raman spectra. Despite the lower discrimination value for the in vivo transcutaneous measurements, which could be explained by the breathing movement and skin influences, our results showed good accuracy in discriminating between normal and cancer breast tissue samples. To support this, the calculated integration areas from the receiver-operating characteristic (ROC) curve yielded 0.86, 0.94, and 1.0 for in vivo transcutaneous, in vivo skin-removed, and ex vivo biopsy measurements, respectively. The feasibility of using HF Raman spectroscopy as a clinical diagnostic tool for breast cancer detection and monitoring is due to no interfering contribution from the optical fiber in the HF Raman region, the shorter acquisition time due to a more intense signal in the HF Raman region, and the ability to distinguish between normal and cancerous tissues.     

Diagnosis of gastric endoscopic biopsies using attenuated total reflectance (ATR) Fourier transform infrared (FT-IR) spectroscopy with entropy weight local-hyperplane k-nearest neighbor based on frequency domain information (EWHFI)

Fourier transform infrared spectroscopy (FT-IR) is an effective diagnostic tool for the characterization of biological molecules that may provide uninvasive, rapid detection of gastric cancer in vivo in the future. FT-IR spectra were obtained from 103 endoscopic biopsy gastric tissue samples that included 20 healthy, 35 chronic atrophic gastritis, 29 chronic superficial gastritis, and 19 gastric cancerous samples. A new approach using entropy weight local-hyperplane k-nearest neighbor based on frequency domain information (EWHFI) is proposed that improves and extends the adaptive weight local-hyperplane k-nearest neighbor (AWHK) approach. EWHFI combines the AWHK algorithm with the fine Fourier transform information in the frequency domain to effectively distinguish similar pathological states. The experimental results show that EWHFI enhances the accuracy, sensitivity, and specificity for the diagnosis of early gastric cancer. The average results of the random classification showed that the EWHFI classifiers for differentiating gastric cancerous, chronic atrophic gastritis, chronic superficial gastritis, and healthy tissues were 95.1%, 86.4%, 88.3%, and 99%, respectively.     

CH-overtone regions as diagnostic markers for near-infrared spectroscopic diagnosis of primary cancers in human pancreas and colorectal tissue

We have investigated the application of near-infrared spectroscopy for detection of human primary pancreatic and colorectal cancers. Spectra from cancerous and normal tissue were collected from a total of 37 surgically resected pancreatic and colorectal patient tissue specimens using a fibre-optic probe. Major spectral differences were observed in the CH-stretching first (6,000–5,400 cm⁻¹) and second overtone (9,000–7,900 cm⁻¹) regions. By use of artificial neural networks, linear discriminant analysis, and cluster analysis as pattern-recognition methods the spectra were classified into cancerous and normal tissue groups with accuracy up to 89%. We also explored differences between the spectra obtained from colorectal and pancreatic tissue. Spectral data from cancerous and normal tissue were classified organ-specifically into four groups with accuracy between 80 and 83%. Our results indicate that CH-overtone regions, besides serving as diagnostic markers for NIR spectroscopic diagnosis of primary human pancreas and colorectal cancers, are also useful for elucidating differences between the spectra obtained from colorectal and pancreatic cancerous tissue.     

Depth profiling of calcifications in breast tissue using picosecond Kerr-gated Raman spectroscopy

Breast calcifications are found in both benign and malignant lesions and their composition can indicate the disease state. Calcium oxalate (dihydrate) (COD) is associated with benign lesions, however calcium hydroxyapatite (HAP) is found mainly in proliferative lesions including carcinoma. The diagnostic practices of mammography and histopathology examine the morphology of the specimen. They can not reliably distinguish between the two types of calcification, which may indicate the presence of a cancerous lesion during mammography. We demonstrate for the first time that Kerr-gated Raman spectroscopy is capable of non-destructive probing of sufficient biochemical information from calcifications buried within tissue, and this information can potentially be used as a first step in identifying the type of lesion. The method uses a picosecond pulsed laser combined with fast temporal gating of Raman scattered light to enable spectra to be collected from a specific depth within scattering media by collecting signals emerging from the sample at a given time delay following the laser pulse. Spectra characteristic of both HAP and COD were obtained at depths of up to 0.96 mm, in both chicken breast and fatty tissue; and normal and cancerous human breast by utilising different time delays. This presents great potential for the use of Raman spectroscopy as an adjunct to mammography in the early diagnosis of breast cancer.     

Spectroscopic diagnosis of colonic dysplasia.

We have developed a method for defining diagnostic algorithms for pathologic conditions based on fluorescence spectroscopy. We apply this method to human colon tissue and show that fluorescence can be used to diagnose the presence or absence of colonic adenoma. This method uses fluorescence excitation-emission matrices (EEM) to identify optimal excitation regions for obtaining fluorescence emission spectra which can be used to differentiate normal and pathologic tissues. In the case of normal and adenomatous colon tissue, these were found to be: 330, 370, and 430 nm +/- 10 nm. At these excitation wavelengths, emission wavelengths for use in diagnostic algorithms are identified from average difference and ratio of the spectra from normal and pathologic tissues. In colon tissue, at 370 nm excitation, 404, 480, and 680 nm were found to be useful emission wavelengths for diagnosing the presence of adenoma in vitro. The basis of colon tissue autofluorescence was investigated using EEM of pure molecules and relevant excitation-emission maxima in the literature.     

Selective sampling using confocal Raman spectroscopy provides enhanced specificity for urinary bladder cancer diagnosis

In recent years, Raman spectroscopy has shown substantive promise in diagnosing bladder cancer, especially due to its exquisite molecular specificity. The ability to reduce false detection rates in comparison to existing diagnostic tools such as photodynamic diagnosis makes Raman spectroscopy particularly attractive as a complementary diagnostic tool for real-time guidance of transurethral resection of bladder tumor (TURBT). Nevertheless, the state-of-the-art high-volume Raman spectroscopic probes have not reached the expected levels of specificity thereby impeding their clinical translation. To address this issue, we propose the use of a confocal Raman probe for bladder cancer diagnosis that can boost the specificity of the diagnostic algorithm based on its suppression of the out-of-focus non-analyte-specific signals emanating from the neighboring normal tissue. In this article, we engineer and apply such a probe, having depth of field of approximately 280?μm, for Raman spectral acquisition from ex vivo normal and cancerous TURBT samples. Using this clinical dataset, a diagnostic algorithm based on principal component analysis and logistic regression is developed. We demonstrate that this approach results in comparable sensitivity but significantly higher specificity in relation to high-volume Raman spectral data. The application of only two principal components is sufficient for the discrimination of the samples underlining the robustness of the algorithm. Further, no discordance between replicate spectra is observed emphasizing the reproducible nature of the current diagnostic assessment. The high levels of sensitivity and specificity achieved in this proof-of-concept study opens substantive avenues for application of a confocal Raman probe during endoscopic procedures related to diagnosis and treatment of bladder cancer.

Spectrally resolved fluorescence imaging of human colonic adenomas

Native fluorescence (autofluorescence) of human tissues can be a valuable source of diagnostic information for detecting premalignant and malignant lesions in the human body. Digital imaging of autofluorescence may be useful for localization of such lesions during endoscopic examinations. Tissue fluorescence of 31 adenomatous polyps obtained from 16 patients has been excited in vitro using the 325 nm line of a He-Cd laser. Digital images of the autofluorescence are recorded in six spectral bands. This study provides new data about the spatial distributions of autofluorescence intensities emitted in different spectral bands by colonic adenomatous lesions and normal colonic mucosa. Areas characterized by autofluorescence intensity lower than in normal mucosa are found for a majority of the polyps under study. The observed patterns of spatial distribution differ for the different spectral bands and for different polypoid lesions. No inverse correlation is found between the emission intensity and the thickness of colonic mucosa. The results indicate the spectral bands most useful for diagnostic applications and demonstrate the complexity of the optical processes involved in shaping both the spectra and intensities of the autofluorescence.     

Identification of Colorectal Cancer Using Near-Infrared Spectroscopy and Adaboost with Decision Stump

Rapid and objective detection of cancer is crucial for successful treatment. Near-infrared (NIR) spectroscopy is a vibrational technique capable of optically probing molecular changes associated with disease. The purpose of this study was to explore NIR spectroscopy for discriminating cancer from normal colorectal tissues. A total of 110 tissue samples from patients who underwent operations were characterized in this study. The popular ensemble technique AdaBoost was used to construct the diagnostic model. A decision stump was used as the weak learning algorithm. Adaboost with decision stump, an ensemble of weak classifiers, was compared with the most suitable single model, a strong classifier. Only the 20 most significant variables were selected as inputs for the model based on measured defined variable importance. Using an independent test set, the single strong classifier provided diagnostic accuracy of 89.1%, sensitivity of 100%, and specificity of 78.6%, whereas the ensemble of weak stumps provided accuracy of 96.3%, sensitivity of 96.3%, and specificity of 96.3% for distinguishing cancer from normal colorectal tissues. Therefore, NIR spectroscopy in combination with AdaBoost with decision stumps has demonstrated potential for rapid and objective diagnosis of colorectal cancer.     

Near-infrared Fluorescence Spectroscopy Detects Alzheimer's Disease In Vitro

The purpose of this study was to investigate whether near-infrared (NIR) fluorescence spectroscopy could be used to detect Alzheimer's disease (AD) by brain tissue autofluorescence. Unfixed temporal cortex specimens from AD cases and age-matched, non-AD controls were frozen at autopsy and then thawed just prior to spectral measurement. Spectra of intrinsic tissue fluorescence induced by 647 nm light were recorded from 650 to 850 nm. We used principal component analysis of the tissue spectra from 17 AD cases and 5 non-AD control cases in a calibration study to establish a diagnostic algorithm. Retrospectively applied to the calibration set, the algorithm correctly classified 23 of 24 specimens. In a prospective study of 19 specimens from 5 AD brains and 2 non-AD control brains, 3 of the 4 control specimens and all AD specimens were correctly diagnosed. Both the excitation light used and the measured brain tissue autofluorescence are at NIR wavelengths that can propagate through skull and overlying tissue. Therefore, our results demonstrate an optical spectroscopic technique that carries direct molecular level information about disease. This is the first step toward a clinical tool that has the potential to be applied to the noninvasive diagnosis of AD in living patients.     

Raman spectroscopy and imaging: applications in human breast cancer diagnosis

The applications of spectroscopic methods in cancer detection open new possibilities in early stage diagnostics. Raman spectroscopy and Raman imaging represent novel and rapidly developing tools in cancer diagnosis. In the study described in this paper Raman spectroscopy has been employed to examine noncancerous and cancerous human breast tissues of the same patient. The most significant differences between noncancerous and cancerous tissues were found in regions characteristic for the vibrations of carotenoids, lipids and proteins. Particular attention was paid to the role played by unsaturated fatty acids in the differentiation between the noncancerous and the cancerous tissues. Comparison of Raman spectra of the noncancerous and the cancerous tissues with the spectra of oleic, linoleic, α-linolenic, γ-linolenic, docosahexaenoic and eicosapentaenoic acids has been presented. The role of sample preparation in the determination of cancer markers is also discussed in this study.     

De novo peptide sequencing by tandem MS using complementary CID and electron transfer dissociation

De novo sequencing of peptides using tandem MS is difficult due to missing fragment ions in the spectra commonly obtained after CID of peptide precursor ions. Complementing CID spectra with spectra obtained in an ion‐trap mass spectrometer upon electron transfer dissociation (ETD) significantly increases the sequence coverage with diagnostic ions. In the de novo sequencing algorithm CompNovo presented here, a divide‐and‐conquer approach was combined with an efficient mass decomposition algorithm to exploit the complementary information contained in CID and ETD spectra. After optimizing the parameters for the algorithm on a well‐defined training data set obtained for peptides from nine known proteins, the CompNovo algorithm was applied to the de novo sequencing of peptides derived from a whole protein extract of Sorangium cellulosum bacteria. To 2406 pairs of CID and ETD spectra contained in this data set, 675 fully correct sequences were assigned, which represent a success rate of 28.1%. It is shown that the CompNovo algorithm yields significantly improved sequencing accuracy as compared with published approaches using only CID spectra or combined CID and ETD spectra.     

In vivo brain tumor demarcation using optical spectroscopy

The applicability of optical spectroscopy for intraoperative detection of brain tumors/tumor margins was investigated in a pilot clinical trial consisting of 26 brain tumor patients. The results of this clinical trial suggest that brain tumors and infiltrating tumor margins (ITM) can be effectively separated from normal brain tissues in vivo using combined autofluorescence and diffuse-reflectance spectroscopy. A two-step empirical discrimination algorithm based on autofluorescence and diffuse reflectance at 460 and 625 nm was developed. This algorithm yields a sensitivity and specificity of 100 and 76%, respectively, in differentiating ITM from normal brain tissues. Blood contamination was found to be a major obstacle that attenuates the accuracy of brain tumor demarcation using optical spectroscopy. Overall, this study indicates that optical spectroscopy has the potential to guide brain tumor resection intraoperatively with high sensitivity.     

Spectral discrimination of benign and malignant prostate tissues--a preliminary report

In this preliminary report, benign (n=8) and malignant (n=5) prostate tissues, in vitro, have been taken through autofluorescence spectroscopy. Employing Stokes' shift spectra and fluorescence emission spectra as tools of analysis, we were able to discriminate the two sets of tissues with sensitivity and specificity in excess of 85%. When the excised prostate chips were scanned with a spatial resolution of 1 mm, the epicenter of malignancy also could be delineated.     

Raman Spectroscopy and Imaging Studies of Human Digestive Tract Cells and Tissues—Impact of Vitamin C and E Supplementation

Cancers of digestive tract such as colorectal cancer (CRC) and gastric cancer (GC) are the most commonly detected types of cancer worldwide and their origin can be associated with oxidative stress conditions. Commonly known and followed antioxidants, such as vitamin C and E, are widely considered as potential anti-cancer agents. Raman spectra have great potential in the biochemical characterization of matter based on the fact that each molecule has its own unique vibrational properties. Raman spectroscopy allows to precisely characterize components (proteins, lipids, nucleic acids). The paper presents the application of the Raman spectroscopy technique for the analysis of tissue samples and cells of the human colon and stomach. The main goal of this study is to show the differences between healthy and cancerous tissues from the human digestive tract and human normal and cancer colon and gastric cell lines. The paper presents the spectroscopic characterization of normal colon cells, CCD-18 Co, in physiological and oxidative conditions and effect of oxidative injury of normal colon cells upon supplementation with vitamin C at various concentrations based on Raman spectra. The obtained results were related to the Raman spectra recorded for human colon cancer cells—CaCo-2. In addition, the effect of the antioxidant in the form of vitamin E on gastric cancer cells, HTB-135, is presented and compared with normal gastric cells—CRL-7869. All measured gastric samples were biochemically and structurally characterized by means of Raman spectroscopy and imaging. Statistically assisted analysis has shown that normal, ROS injured and cancerous human gastrointestinal cells can be distinguished based on their unique vibrational properties. ANOVA tests, PCA (Principal Component Analysis) and PLSDA (Partial Least Squares Discriminant Analysis) have confirmed the main role of nucleic acids, proteins and lipids in differentiation of human colon and stomach normal and cancer tissues and cells. The conducted research based on Raman spectra proved that antioxidants in the form of vitamin C and E exhibit anti-cancer properties. In consequence, conducted studies proved that label-free Raman spectroscopy may play an important role in clinical diagnostic differentiation of human normal and cancerous gastrointestinal tissues and may be a source of intraoperative information supporting histopathological analysis.     

Optical diagnosis of peritoneal metastases by infrared microscopic imaging

Fourier transform infrared (FTIR) spectroscopy is nowadays widely accepted as a technique with high potential for diagnosis of cancerous tissues. This study presents an example of the investigation of peritoneal metastases by FTIR microimaging. Peritoneal malignancies are generally secondary localizations of primary visceral cancers such as ovarian, stomach or colon cancers. By analysing simultaneously both formalin-fixed paraffin-embedded and frozen specimens, we examined malignant and non-malignant (i.e. fibrotic and cicatricial) peritoneal lesions. Paraffin-embedded tissues were analysed without any previous dewaxing. Multivariate statistical approaches, based on the classification of infrared data by hierarchical cluster analysis, allowed the discrimination of these various samples. Microimaging also permits the revelation of the heterogeneity of the tissue: it was possible to localize precisely the cancerous areas, and to distinguish, on the basis of their spectral signatures, the peritumoral neighbouring connective tissue close to the carcinomatous areas from the connective tissue distant from the cancerous areas. These spectral differences could be useful as complementary information to study molecular changes associated with the malignancy.