Multivariate Classification Techniques Research Articles

Metabolomic profiling of Fiore Sardo cheese: Investigation of the influence of thermal treatment and ripening time using univariate and multivariate classification techniques

The effect of different sub-pasteurization heat treatments and different ripening times was investigated in this work. The metabolite profiles of 95 cheese samples were analyzed using GC–MS in order to determine the effects of thermal treatment (raw milk, 57 °C and 68 °C milk thermization) and ripening time (105 and 180 days). ANOVA test on GC–MS peaks complemented with false discovery rate correction was employed to identify the compounds whose levels significantly varied over different ripening times and thermal treatments. The univariate t-test classifier and Partial Least Square Discriminant Analysis (PLS-DA) provided acceptable classification results, with an overall accuracy in cross-validation of 76% for the univariate model and 72% from the PLS-DA. The metabolites that mostly changed with ripening time were amino acids and one endocannabinoid (i.e., arachidonoyl amide), while compounds belonging to the classes of biogenic amines and saccharides resulted in being strongly affected by the thermization process.

Recognition of breast cancer subtypes using FTIR hyperspectral data

Fourier-transform infrared spectroscopy (FTIR) is a powerful, non-destructive, highly sensitive and a promising analytical technique to provide spectrochemical signatures of biological samples, where markers like carbohydrates, proteins, and phosphate groups of DNA can be recognized in biological micro-environment. However, method of measurements of large cells need an excessive time to achieve high quality images, making its clinical use difficult due to speed of data-acquisition and lack of optimized computational procedures. To address such challenges, Machine Learning (ML) based technologies can assist to assess an accurate prognostication of breast cancer (BC) subtypes with high performance. Here, we applied FTIR spectroscopy to identify breast cancer subtypes in order to differentiate between luminal (BT474) and non-luminal (SKBR3) molecular subtypes. For this reason, we tested multivariate classification technique to extract feature information employing three-dimension (3D)-discriminant analysis approach based on 3D-principle component analysis-linear discriminant analysis (3D-PCA-LDA) and 3D-principal component analysis-quadratic discriminant analysis (3D-PCA-QDA), showing an improvement in sensitivity (98%), specificity (94%) and accuracy (98%) parameters compared to conventional unfolded methods. Our results evidence that 3D-PCA-LDA and 3D-PCA-QDA are potential tools for discriminant analysis of hyperspectral dataset to obtain superior classification assessment.

Enhancing Skin Lesion Detection: A Multistage Multiclass Convolutional Neural Network-Based Framework.

The early identification and treatment of various dermatological conditions depend on the detection of skin lesions. Due to advancements in computer-aided diagnosis and machine learning approaches, learning-based skin lesion analysis methods have attracted much interest recently. Employing the concept of transfer learning, this research proposes a deep convolutional neural network (CNN)-based multistage and multiclass framework to categorize seven types of skin lesions. In the first stage, a CNN model was developed to classify skin lesion images into two classes, namely benign and malignant. In the second stage, the model was then used with the transfer learning concept to further categorize benign lesions into five subcategories (melanocytic nevus, actinic keratosis, benign keratosis, dermatofibroma, and vascular) and malignant lesions into two subcategories (melanoma and basal cell carcinoma). The frozen weights of the CNN developed-trained with correlated images benefited the transfer learning using the same type of images for the subclassification of benign and malignant classes. The proposed multistage and multiclass technique improved the classification accuracy of the online ISIC2018 skin lesion dataset by up to 93.4% for benign and malignant class identification. Furthermore, a high accuracy of 96.2% was achieved for subclassification of both classes. Sensitivity, specificity, precision, and F1-score metrics further validated the effectiveness of the proposed multistage and multiclass framework. Compared to existing CNN models described in the literature, the proposed approach took less time to train and had a higher classification rate.

A lateralized alpha-band marker of the interference of exogenous attention over endogenous attention.

Scientists differentiate top-down from bottom-up visuospatial attention processes. While bottom-up attention is rapidly engaged by emerging demands from the environment, top-down attention in contrast reflects slow voluntary shifts of spatial attention. Several lines of research substantiate the idea that top-down and bottom-up attentional processes involve distinct functional systems. An increasing number of studies, however, argue that both attention systems share brain processing resources. The current study examines how salient visual events that engage bottom-up processes interfere with top-down attentional processes. Using neurophysiological recordings and multivariate pattern classification techniques, the authors show that these patterns of interference occur within the alpha-band of neural activity (8-12Hz), which implies that bottom-up and top-down attention processes share this narrow-band frequency brain resource. The results further demonstrate that patterns of alpha-band activity explains, in part, the interference between top-down and bottom-up attention at the behavioral level.

A Systematic Review of UAV Applications for Mapping Neglected and Underutilised Crop Species’ Spatial Distribution and Health

Timely, accurate spatial information on the health of neglected and underutilised crop species (NUS) is critical for optimising their production and food and nutrition in developing countries. Unmanned aerial vehicles (UAVs) equipped with multispectral sensors have significantly advanced remote sensing, enabling the provision of near-real-time data for crop analysis at the plot level in small, fragmented croplands where NUS are often grown. The objective of this study was to systematically review the literature on the remote sensing (RS) of the spatial distribution and health of NUS, evaluating the progress, opportunities, challenges, and associated research gaps. This study systematically reviewed 171 peer-reviewed articles from Google Scholar, Scopus, and Web of Science using the PRISMA approach. The findings of this study showed that the United States (n = 18) and China (n = 17) were the primary study locations, with some contributions from the Global South, including southern Africa. The observed NUS crop attributes included crop yield, growth, leaf area index (LAI), above-ground biomass (AGB), and chlorophyll content. Only 29% of studies explored stomatal conductance and the spatial distribution of NUS. Twenty-one studies employed satellite-borne sensors, while only eighteen utilised UAV-borne sensors in conjunction with machine learning (ML), multivariate, and generic GIS classification techniques for mapping the spatial extent and health of NUS. The use of UAVs in mapping NUS is progressing slowly, particularly in the Global South, due to exorbitant purchasing and operational costs, as well as restrictive regulations. Subsequently, research efforts must be directed toward combining ML techniques and UAV-acquired data to monitor NUS’ spatial distribution and health to provide necessary information for optimising food production in smallholder croplands in the Global South.

A Deep Learning Technique to Improve Road Maintenance Systems Based on Climate Change

Road maintenance systems (RMS) are crucial for maintaining safe and efficient road networks. The impact of climate change on road maintenance systems is a concern as it makes them more susceptible to weather events and subsequent damages. To tackle this issue, we propose an RMSDC (Road Maintenance Systems Using Deep Learning and Climate Adaptation) technique to improve road maintenance systems based on Deep learning and Climate Adaptation. RMSDC aims to use the multivariate classification technique and divides the dataset into training and test datasets. The RMSDC combines Convolutional Long Short-Term Memory (ConvLSTM) techniques with road weather information and sensor data. However, in emerging nations, the effects of climate change are already apparent, which makes road networks particularly susceptible to extreme weather, floods, and landslides. Therefore, climate adaptation of road networks is essential, especially in developing nations with limited financial resources. To address this issue, we propose an intelligent and effective RMSDC that utilizes deep learning algorithms based on climate change predictions. The ConvLSTM block effectively captures the relationship between input features over time to calculate the root-mean deviation (RMSD). We evaluate RMSDC performance against frameworks for downscaling climate variables using two metrics: root-mean-square error (RMSE) and mean absolute difference. Through real evaluations, RMSDC consistently outperforms approaches with a reduced RMSE of 0.26. These quantitative results highlight how effective RMSDC is in addressing maintenance systems on road networks leading to proactive road maintenance strategies that enhance traffic safety, reduce costs, and improve environmental sustainability.

MBFRDH: Design of a Multimodal Bioinspired Feature Representation Deep Learning Model for Identification of Heart-Diseases

Electrocardiograms (ECGs) are generated by checking different beating patterns of heart, and are widely used for identification of multiple heart-related issues. Existing deep learning models that are proposed for ECG analysis are either highly complex, or showcase lower scalability when applied to clinical scans. To overcome these issues, this text proposes design of a novel multimodal bioinspired feature representation deep learning model for identification of heart-diseases. The proposed model initially collects large-scale ECG datasets, and extracts Fourier, Cosine, iVector, Gabor, and Wavelet components. These components are given to a Grey Wolf Optimization (GWO) based feature selection model, which assists in identification of high-inter-class variance feature sets. This is done via modelling a variance-based fitness function and fusing it with an Iterative Learning Model (ILM) that use feedback-accuracy levels for optimization of selected feature sets. The extracted features are used to incrementally train a custom 1D Binary-Augmented Convolutional Neural Network (1D BACNN) that can be trained for multiclass scenarios. The BACNN Model is trained individually for each of the heart diseases. Each BACNN categorizes input ECG samples between ‘Normal’, and ‘Heart-Disease’ categories. Due to use of this binary-type classification, the proposed model is able to achieve a consistent 99.9% accuracy for multiple heart disease sets, which is found to be higher than most of the existing multiclass techniques. The model was tested for Angina, Arrhythmia, Valve disease, and Congenital heart conditions, and was observed to achieve 3.5% higher precision, 4.9% higher accuracy, with 1.2% increase is computational delay, which makes it highly suitable for real-time clinical use cases.

A model based on a multivariate classification for assessing impacts on water quality in a DOCE river watershed after the Fundão tailings dam failure

The main purpose of this study was to build multivariate classification models using water quality monitoring data for the hydrographic basin of the Gualaxo do Norte River, Minas Gerais state, Brazil, which was impacted in 2015 by the rupture of a containment structure for iron ore tailings. A total of 27 points were evaluated, covering areas affected and unaffected by the disaster, with monitoring of chemical, physical, and microbiological variables during the period from July 2016 to June 2017. Multivariate classification techniques were applied to the data, with the aim of developing models to determine when the impacted locations would present characteristics equivalent to those existing prior to the rupture. Classification models constructed using PLS-DA and LDA were able to predict three classes: unaffected main river, affected main river, and tributaries. The first technique was able to clearly differentiate the three classes for the data evaluated, achieving averages corresponding to 90% accuracy. The second method was consistent with the first, identifying the chloride content, conductivity, turbidity, and alkalinity as discriminatory variables, among those monitored, with the relationships among the parameters being coherent with the environmental conditions of the region. The model, with a correct classification rate of 91.67%, enabled identification of the behavior of new samples, using only these easily measured variables. In summary, application of the multivariate statistical tools allowed the development of models capable of providing information about the recovery process of an ecosystem impacted by the greatest environmental disaster to have occurred in Brazil.

Linear discriminant analysis on electrocardiogram achieved classification of cardiac involvement status in amyloid light-chain amyloidosis

ObjectivesCardiac amyloidosis (CA) is the most crucial determinant of amyloid light-chain (AL) amyloidosis patients' prognosis. We attempted cardiac involvement prediction by 12‑lead electrocardiograph (ECG) and echocardiography (UCG) in AL amyloidosis patients. Materials and methodsFifty patients with histologically confirmed AL amyloidosis underwent gadolinium-enhanced magnetic resonance imaging (Gd-MRI), and CA was assessed using late gadolinium enhancement. ECG and UCG parameters were measured on admission. Fisher's linear discriminant analysis was used to create a model for predicting CA using the ECG and UCG parameters. ResultsPrediction by five ECG parameters [QTc(B), QRS-T-angle, III-QRS, aVF-QRS, and V3-R] showed the best performance. Average sensitivity and specificity in the modeling sets, utilizing a linear discriminator based on these five variables, were 99.2 % and 96.8 % and in validation sets, 94.2 % and 90.3 %, respectively. In addition, we tested this model on an additional 26-patient cohort and survival analysis using the Kaplan-Meier method, and significant differences between CA positively predicted and negatively predicted patients were observed. ConclusionHere, we suggest the application of a condensed classical multivariate statistical technique for the diagnosis of CA. It can be used as a guide to invasive endomyocardial biopsy for those in whom Gd-MRI is contraindicated and as a guide for repeat Gd-MRI in follow-up of AL amyloidosis.

Multivariate Statistical Process Control and Classification Applied on Prostate Cancer Screening

Introduction: We report in this study the results of analyzing biomarkers in blood samples with two objectives; i) as an approach for screening patients by use of Multivariate Statistical Process Control (MSPC); ii) Compare various classification methods with the purpose of diagnosing prostate cancer. Methods: We applied Principal Component Analysis (PCA) with statistical limits for outlier detection. Various splits of the data into training and test sets were chosen to evaluate the performance of classification methods as a function of the training/test sample ratio. Results: MSPC based on 12 analytes in blood samples was shown to outperform the traditional biomarker criterion: the level of the analyte Prostate-Specific Antigen (PSA), in screening for prostate cancer. The performance of different multivariate classification techniques for classifying which of the patients in a clinical pathway for prostate cancer have malignant tumors showed that the basic method Linear Discriminant Analysis (LDA) and classification trees gave similar results, whereas adaboost gave a higher specificity but lower sensitivity. Conclusion: The accuracy, especially the sensitivity, does not justify any clinical use of the applied classification methods with the available biomarkers. Additional medical information about the patients might enhance the accuracy with the purpose of identifying benign and malignant tumors.

Classification of instant coffees based on caffeine content and roasting degree using NIR spectrometry and multivariate analysis

Instant coffee is a beverage obtained from the dehydration of roasted coffee extract. Its quality depends on several factors, such as the type of bean and grinding, drying decaffeination, and roasting degree processes. In this work, NIR spectrometry and multivariate classification tools were combined to develop, for the first time, methods for authenticating and checking the compliance of decaffeinated instant coffees in order to ensure consumer food safety, as well as discriminating between regular instant coffees according to the roasting degree (traditional and extra-strong) for monitoring the production chain of regular instant coffees. First, one-class methods were built and validated to authenticate decaffeinated instant coffees and to in order to alert caffeine-sensitive consumers to potential risks. For this purpose, two near-infrared (NIR) instruments (benchtop and portable) and Data-Driven Soft Independent Modeling of Class Analogy (DD-SIMCA) and One-Class Partial Least Squares (OC-PLS) were used. After Partial Least Squares – Discriminant Analysis (PLS-DA), Successive Projections Algorithm for Interval Selection in Partial Least Squares – Discriminant Analysis (iSPA-PLS-DA), and the same NIR instruments were used to build and validate multiclass methods to discriminate regular instant coffees according to the roasting degree (traditional and extra-strong). The result for the classification of decaffeinated and roasting degree instant coffees was 100% and 98% correct classification rate using DD-SIMCA and iSPA-PLS-DA, respectively. Therefore, the combination of NIR spectroscopy and multivariate classification techniques proved valuable and fast tools for checking the quality of instant coffees.

Fourier transform infrared (FTIR) spectroscopy as a screening tool for osteosarcopenia in community-dwelling older women.

Osteosarcopenia is a complex geriatric syndrome characterized by the presence of both sarcopenia and osteopenia/osteoporosis. This condition increases rates of disability, falls, fractures, mortality and mobility impairments in older adults. The purpose of this study was to analyze the Fourier Transform Infrared (FTIR) spectroscopy diagnostic power for osteosarcopenia in community-dwelling older women (n = 64, 32 osteosarcopenic and 32 non-osteosarcopenia). FTIR is a fast and reproducible technique highly sensitive to biological tissues and a mathematical model was created using multivariate classification techniques that denoted the graphic spectra of the molecular groups. Genetic algorithm and support vector machine regression (GA-SVM) was the most feasible model, achieving 80.0% of accuracy. GA-SVM identified 15 wavenumbers responsible for class differentiation, in which several amino acids (responsible for the proper activation of the mammalian target of rapamycin) and hydroxyapatite (an inorganic bone component) were observed. Imaging tests and a low availability of instruments that allow the observation of osteosarcopenia involves high health costs for patients and restrictive indications. Therefore, FTIR can be used to diagnose osteosarcopenia due to its efficiency, low cost and to enable early detection in the geriatric services, contributing to advances in science and technology that are potential "conventional" methods in the future.

Multiple adulterants detection in turmeric powder using Vis-SWNIR hyperspectral imaging followed by multivariate curve resolution and classification techniques

In the present contribution, visible and short wavelengths of near infrared hyperspectral imaging (Vis-SWNIR-HSI) combined with different chemometric techniques is proposed as a novel technique for turmeric authentication and multiple adulterants (corn flour, rice flour, starch, wheat flour, and zedoary) detection. In this regard, twenty-three samples of turmeric were collected as whole rhizomes or powdered from seven countries and then their VIS-SWNIR hyperspectral images were obtained in 400–1000 nm using SPECIM IQ HSI device. Two multivariate resolution techniques of multivariate curve resolution-alternating least squares (MCR-ALS) and mean-field independent component analysis (MF-ICA) were used to extract pure spatial and spectral profiles of the components, and their results were compared by projecting their solutions in the area of feasible solutions (AFSs). Then, the distribution maps of turmeric component obtained using MCR-ALS and MF-ICA were used for authentication. Principal component analysis (PCA) was used to find the pattern of authentic samples using their distribution maps. Additionally, data-driven soft independent modeling of class analogy (DD-SIMCA) was employed to find boundary between authentic and adulterated turmeric samples. On this matter, good class modelling results with sensitivity of 95 % and specificity of 100 % were obtained. Finally, partial least squares-discriminant analysis (PLS-DA) was utilized for discrimination of the of adulternats and their in binary and multi-class classification modes. On this matter, PLS-DA accuracies were acceptable for binary and mixed samples which confirmed the validity of the proposed method.

Assessment of meat authenticity using portable Fourier transform infrared spectroscopy combined with multivariate classification techniques

In the present contribution, the feasibility of portable Fourier transform infrared spectroscopy (FTIR) combined with multivariate classification techniques is assessed for classification of minced beef, lamb, chicken and pork samples. In this regard, both attenuated total reflectance-FTIR (ATR-FTIR) and diffuse reflectance-FTIR (DR-FTIR) methods are evaluated. First, principal component analysis (PCA) was used for exploring FT-IR spectra of four meat species to find similarities and dissimilarities among samples. Additionally, one-class classification (OCC) was utilized as a new approach for halal meat species certification. For OCC, two scenarios were defined: (i) 100% correct classification for pork, and (ii) a most favorable overall classification rate for all species investigated simultaneously. With the OCC approach, both ATR and DR methods were found to produce high false-positive scores in scenario (i), whilst the DR method scored the best in scenario (ii) with an overall score of 89% correct classification. In the next step, partial least squares-discriminant analysis (PLS-DA) and support vector machine (SVM) with radial basis function (RBF) as kernel function were evaluated for meat speciation. On this matter, SVM showed better classification performance in terms of total accuracy for both ATR-FTIR (98%) and DR-FTIR (100%) datasets over PLS-DA (90% and 98%, respectively).The promising results of both portable ATR-FTIR and DR-FTIR combined with OCC approach and discriminant analysis indicated for the first time their use as successful non-destructive, cost-effective and rapid routine screening methods for on-site analysis of meat speciation and halal meat species certification which could be useful for quality control officers to manage and control meat authenticity at various stages of the supply chain.

Geochemical and Spatial Distribution of Topsoil HMs Coupled with Modeling of Cr Using Chemometrics Intelligent Techniques: Case Study from Dammam Area, Saudi Arabia.

Unconsolidated earthen surface materials can retain heavy metals originating from different sources. These metals are dangerous to humans as well as the immediate environment. This danger leads to the need to assess various geochemical conditions of the materials. In this study, the assessment of topsoil materials’ contamination with heavy metals (HMs) was conducted. The material’s representative spatial samples were taken from various sources: agricultural, industrial, and residential areas. The materials include topsoil, eolian deposits, and other unconsolidated earthen materials. The samples were analyzed using the ICP-OES. The obtained results based on the experimental procedure indicated that the average levels of the heavy metals were: As (1.21 ± 0.69 mg/kg), Ba (110.62 ± 262 mg/kg), Hg (0.08 ± 0.18 mg/kg), Pb (6.34 ± 14.55 mg/kg), Ni (8.95 ± 5.66 mg/kg), V (9.98 ± 6.08 mg/kg), Cd (1.18 ± 4.33 mg/kg), Cr (31.79 ± 37.9 mg/kg), Cu (6.76 ± 12.54 mg/kg), and Zn (23.44 ± 84.43 mg/kg). Subsequently, chemometrics modeling and a prediction of Cr concentration (mg/kg) were performed using three different modeling techniques, including two artificial intelligence (AI) techniques, namely, generalized neural network (GRNN) and Elman neural network (Elm NN) models, as well as a classical multivariate statistical technique (MST). The results indicated that the AI-based models have a superior ability in estimating the Cr concentration (mg/kg) than MST, whereby GRNN can enhance the performance of MST up to 94.6% in the validation step. The concentration levels of most metals were found to be within the acceptable range. The findings indicate that AI-based models are cost-effective and efficient tools for trace metal estimations from soil.

Maturation of large-scale brain systems over the first month of life.

The period immediately after birth is a critical developmental window, capturing rapid maturation of brain structure and a child's earliest experiences. Large-scale brain systems are present at delivery, but how these brain systems mature during this narrow window (i.e. first weeks of life) marked by heightened neuroplasticity remains uncharted. Using multivariate pattern classification techniques and functional connectivity magnetic resonance imaging, we detected robust differences in brain systems related to age in newborns (n = 262; R2 = 0.51). Development over the first month of life occurred brain-wide, but differed and was more pronounced in brain systems previously characterized as developing early (i.e. sensorimotor networks) than in those characterized as developing late (i.e. association networks). The cingulo-opercular network was the only exception to this organizing principle, illuminating its early role in brain development. This study represents a step towards a normative brain "growth curve" that could be used to identify atypical brain maturation in infancy.

Vis-NIR hyperspectral imaging coupled with independent component analysis for saffron authentication

In the present contribution, visible-near infrared hyperspectral imaging (Vis-NIR-HSI) combined with a novel chemometric approach based on mean-filed independent component analysis (MF-ICA) followed by multivariate classification techniques is proposed for saffron authentication and adulteration detection. First, MF-ICA was used to exploit pure spatial and spectral profiles of the components. Then, principal component analysis (PCA) and hierarchical cluster analysis (HCA) were used to find patterns of authentic samples based on their distribution maps. Then, detection of five common plant-derived adulterants of saffron including safflower, saffron style, calendula, rubia and turmeric were investigated. For this purpose, partial least squares-discriminant analysis (PLS-DA) for supervised classification to find a boundary between authentic and adulterated saffron samples. Classification accuracies for all models for calibration and prediction sets were 100 %. Finally, a mixed dataset was prepared and analyzed using the proposed strategy which again 100 % of accuracies for calibration and prediction sets were obtained. At the end, data driven soft independent modelling of class analogy (dd-SIMCA) was used to evaluate model for class modeling. Sensitivity was 95% for authentic class and specificities for all adulterants were 100%.

Detection of fabricated eggs using Fourier transform infrared (FT-IR) spectroscopy coupled with multivariate classification techniques

Fabricated eggs can lead to many severe human health problems due to the consumption of the harmful chemicals used to make them. Therefore, developing technology for detecting fabricated eggs is of high priority in terms of food safety. Fourier transform infrared (FT-IR) spectroscopy is widely used for non-destructive detection applications for its capacity to rapidly assess targets via optical properties. In this study, one-way analysis of variance (ANOVA) and principal component analysis (PCA) were performed to interpret FT-IR spectral data related to the chemical composition of fabricated eggs. As a result, particular wavenumbers were observed that were highly sensitive to the external (eggshell) and internal (albumen and yolk) constituents of fabricated and real chickens eggs. Based on the spectral interpretation, multivariate models using partial least-squares discriminant analysis (PLS-DA), partial least-squares regression (PLSR), support vector machine (SVM), and support vector regression (SVR) were built to detect the constituents of fabricated eggs. The PLS-DA and SVM models classified fabricated eggshells and fabricated albumen with 100% accuracy. Moreover, the PLSR and SVR predictive values exhibited coefficients of determination (Rp2) of 0.99. Hence, this study demonstrates that FT-IR spectroscopy combined with multivariate models can be potentially used as an efficient method to detect both surface and internal constituents of fabricated eggs.

Untargeted metabolomics based on nuclear magnetic resonance spectroscopy and multivariate classification techniques for identifying metabolites associated with breast cancer patients

In this study, multivariate classification techniques combined with proton nuclear magnetic resonance (1HNMR) spectroscopy is proposed to identify breast cancer biomarkers that can precisely distinguish between healthy control and breast cancer (BC) patients. In this regard, first optimizing the metabolite extraction procedure was performed using Box-Behnken design (BBD). Then, data-driven soft independent modeling of class analogy (DD-SIMCA) model and partial least squares-discriminant analysis (PLS-DA) were successfully utilized for separating healthy from BC patient's classes. On this matter, both DD-SIMCA and PLS-DA models could successfully distinguish the healthy class from the BC class with the model's sensitivity and specificity of 100%. Variable importance in projection (VIP) method was implemented to detect significant metabolites. Based on significant variables, 13 significant metabolites (e.g., lactic acid, cysteine, and glucose) were detected as the influential factors for this discrimination. Also, a heat map revealing the trend in metabolites levels was depicted and altered metabolic pathways were detected.

Non-invasive identification of textile fibres using near-infrared fibre optics reflectance spectroscopy and multivariate classification techniques

The identification of textile fibres from cultural property provides information about the object's technology. Today, microscopic examination remains the preferred method, and molecular spectroscopies (e.g. Fourier transform infrared (FTIR) and Raman spectroscopies) can complement it but may present some limitations. To avoid sampling, non-invasive fibre optics reflectance spectroscopy (FORS) in the near-infrared (NIR) range showed promising results for identifying textile fibres; but examining and interpreting numerous spectra with features that are not well defined is highly time-consuming. Multivariate classification techniques may overcome this problem and have already shown promising results for classifying textile fibres for the textile industry but have been seldom used in the heritage science field. In this work, we compare the performance of two classification techniques, principal component analysis–linear discrimination analysis (PCA-LDA) and soft independent modelling of class analogy (SIMCA), to identify cotton, wool, and silk fibres, and their mixtures in historical textiles using FORS in the NIR range (1000–1700 nm). We built our models analysing reference samples of single fibres and their mixtures, and after the model calculation and evaluation, we studied four historical textiles: three Persian carpets from the nineteenth and twentieth centuries and an Italian seventeenth-century tapestry. We cross-checked the results with Raman spectroscopy. The results highlight the advantages and disadvantages of both techniques for the non-invasive identification of the three fibre types in historical textiles and the influence their vicinity can have in the classification.