Search

Quality assessment of milk which is a comprehensive source of nutrients for humans and an important raw material for other dairy products is required in the dairy industry. Rapid, cost-effective, and non-destructive spectroscopic techniques are more preferable than classic wet chemistry approaches for milk analysis. The objective of this work was to review the prediction of milk composition including macronutrients such as fat, protein and lactose and micronutrients such as fatty acids and vitamins using multivariate chemometric modelling of Near Infrared (NIR), Mid Infrared (MIR), fluorescence, and Raman spectral data and data fusion approaches. Literature sources describing spectroscopic analysis of milk samples and the application of multivariate data analysis methods are outlined in this literature review. In addition, the importance of data fusion strategies employed for combing different spectroscopic techniques are reviewed to evaluate their potential to improve the accuracy of the prediction models developed. Recent research studies have demonstrated that the use of data fusion strategies improves the performance of milk composition prediction models developed. [ABSTRACT FROM AUTHOR]

Wheat variety identification is an essential component of the official inspection of wheat grains to determine their quality and trade value. Traditionally, wheat class identification requires deep knowledge about the appearance of plants and kernels, their history and distribution. Therefore, there is an urgent need for an automatic technology to standardize the nomenclature of wheat varieties to enable both growers and producers to identify their grains practically. The present study investigated the feasibility of Visible-Near Infrared (Vis-NIR) and Short-Wave Infrared (SWIR) spectral imaging to identify wheat classes. Both sides of the wheat kernel as single kernel measurements and bulk grains were used to acquire hyperspectral data and vertical and horizontal data concatenation was implemented to explore performance differences. In addition, the classification of bulk kernels from single kernel data was investigated. Spectral data was pre-treated by standard normal variate (SNV), Savitzky-Golay first (SG-1) and second derivatives (SG-2) and linear discriminant analysis (LDA), support vector machine (SVM) and artificial neural network (ANN) were applied as the classification algorithms. Furthermore, feature selection was utilised using the minimum redundancy maximum relevance (mRMR) algorithm to investigate the performance difference between data concatenation and feature selection. The results showed that ventral (up) side data showed better classification performances than reverse (down) side data, while Vis-NIR data achieved higher classification accuracies than SWIR data. However, the best classification performance for single kernels was obtained by LDA-SNV using up and down data in the Vis-NIR and SWIR regions vertically and horizontally concatenated with an accuracy of 93.72% for 10-fold cross-validation and 94.93% for test sets. Models based on a hundred features did not achieve the accuracy of models based on concatenated data. Moreover, classification performances of bulk samples were higher than single kernels, which achieved 100% accuracy for both cross-validation and test sets. The study demonstrates that spectral imaging has a high potential to identify wheat classes non-destructively. [Display omitted] • Hyperspectral imaging is evaluated for classification of 37 wheat varieties. • Vis-NIR more accurate than SWIR for kernel classification. • Models built on bulk samples more accurate than single kernel models. • Data concatenation provides more robust results. [ABSTRACT FROM AUTHOR]

Near infrared hyperspectral imaging was used to predict the particulate matter (PM) and gaseous emissions of pellets produced from pinewood with varying moisture contents. PM 10 and SO 2 emissions were successfully determined using partial least squares regression coupled with spectral data of the pellets. The resulting models had R2 values of 0.74 and 0.75 and RMSEPs of 0.29 mg m−3 and 2.01 mg m−3 for the PM 10 and SO 2 emissions, respectively. Correlations for the remaining emissions factors were also determined. The product of the feed rate and flue gas temperature was found to correlate with the PM 1 , TSP and CO emissions with Pearson's r values of 0.48, 0.54 and 0.74, respectively. PM 2.5 and PM 0 emissions correlated with the flue gas temperature (r = 0.48) and bulk density (r = 0.62), respectively. Gaseous emissions of NO x correlated with a Pearson's r–value of 0.57 to the product of the bulk density and the flue gas temperature. The results indicate that near infrared spectroscopy can be used to predict the PM 10 and SO 2 emissions of a single biomass type sample set. Correlations between physical parameters of the pellets and emissions factors can be used for approximate predictions of emissions from wood pellets. Highlights • Near infrared hyperspectral imaging used for predicting emissions of pellets. • PM 10 and SO 2 emissions successfully determined using hyperspectral imaging. • Correlations for PM 1 , TSP and CO emissions determined. • PM 0 emissions correlated with the bulk density (r = 0.62). • A correlation for the NO x emissions was found with a Pearson's r value of 0.57. [ABSTRACT FROM AUTHOR]

Abstract This work aims to investigate the emergence of aggregates caused by redundant plasticizers in the protein matrix of casein based biopolymers using chemical imaging techniques. Near infrared (NIR) images (950–1671 nm) were first acquired and the spatial variations on macroscale with a pixel size of 0.4 mm × 0.5 mm were visualized. The introduction of plasticizers resulted in a strong hydrogen bonding matrix in the protein polymeric film as evidenced by analysis of Fourier transform near infrared (FT-NIR) spectral profiles in the range of 7500–4000 cm−1. Attenuated total reflectance Fourier transform infrared (ATR-FTIR) images (4000–650 cm−1) coupled with principal components analysis (PCA) and multivariate curve resolution alternating least squares (MCR-ALS) analysis suggested the existence of sorbitol re-crystallization after 5 months storage in the ambient condition. Raman images with a higher pixel size of 1.2 µm × 1.2 µm indicated an uneven film surface caused by sorbitol migration and re-crystallization. A partial least squares (PLS) regression model was developed to predict plasticizer concentration based on the mean spectra of FT-NIR hypercubes, producing coefficient of determination in calibration ( R 2 Cal) of 0.93, cross-validation ( R 2 CV) of 0.92 and prediction ( R 2 P) of 0.89. Visualization of aggregates in the image field was obtained by applying the developed PLS model in a pixel-wise manner using single-element and array detectors. The combined information from NIR and FT-NIR evidenced the occurrence of high plasticizer-concentrated regions in the film sample, while the combined information from FT-NIR and ATR-FTIR further confirms the phenomenon of sorbitol re-crystallization. Graphical abstract fx1 Highlights • Plasticizer led to strong hydrogen bonding matrix as revealed by FT-NIR and ATR-FTIR; • ATR-FTIR and Raman imaging suggested the existence of sorbitol re-crystallization; • Visualization of aggregates was obtained by applying PLS model on FT-NIR images; • Combined NIR and FT-NIR information proves the emergence of plasticizer aggregates; • Combined FT-NIR and ATR-FTIR information confirms sorbitol re-crystallization. [ABSTRACT FROM AUTHOR]

Although the ingestion of microplastics by fish has been investigated extensively over the last 50 years, the quality of these ingestion studies is questionable due to the lack of standard methods for sampling, extraction, and identification of microplastics. This review aims to provide reliable results of microplastic occurrence in fish sourced from 104 published articles through a rigorous assessment of the data quality. A scoring system was performed to assess ten quality assurance/quality control criteria including sampling methods, sample size, sample processing and storage, laboratory preparation, clean air conditions, negative controls, positive controls, target component, sample treatment and polymer identification. We found that 59 out of 104 articles had scores ≤10, and only two studies are identified with scores higher than 15, demonstrating the urgent need to improve the quality and reliability of fish digestion studies. Particularly, five criteria, namely, the positive controls, sampling method, clean air conditions, sample size, and polymer identification with an average score of less than 1 need attention and improvement in future studies. This review also revealed that the 104 studies considered here originated mainly from five countries, i.e., China (23%), Iran (8%), Thailand (5%), Turkey (4%), and Indonesia (4%). Among 104 studies, the major polymer type found was polyethylene, with a size range of 0–1 mm. The most frequently reported colour and shape of microplastics found in the fish were blue and fibre, respectively. [ABSTRACT FROM AUTHOR]

This present work aims to provide new insights into the spectral variations associated with water and plasticizers in the protein matrix of casein based biopolymers by conducting time series hyperspectral chemical imaging (HCI). Near infrared (NIR) hyperspectral images (880–1720 nm) of sodium caseinate film were acquired at different periods during film forming process. Second derivative spectra revealed a red shift (from lower to higher wavelength) in the region of 1410–1492 nm during drying, indicating that hydrogen bond distribution converted from free toward associated hydroxyl groups. The introduction of plasticizers resulted in a strong hydrogen bonding matrix in the caseinate polymeric film as evidenced by analysis of second derivative spectra and curve fitting results. Moisture content (MC) of film samples were predicted with coefficient of determination in prediction ( R P r e d 2 ) of 0.96 and 0.97, and root mean square error of prediction (RMSEP) of 0.067 and 0.056 by conducting partial least squares (PLS) and support vector machine (SVM), respectively. SVM model was further used to visualize water distribution within the film samples over drying time and the time-evolution of water migration into a dried film. A novel time series spectral angle mapper (SAM) was also proposed to elucidate water penetration pattern in film material. Time series principal components analysis (PCA) and K-means analysis exhibited temporal variation of film samples during drying steps of film formation. The results clearly demonstrated the occurrence of strong hydrogen-bonded aggregates in the film samples, which was more likely due to the congregation of redundant plasticizers. The overall results confirmed that time series NIR HCI is a powerful and versatile tool to investigate water and plasticizer interactions in the protein-based polymeric film. [ABSTRACT FROM AUTHOR]

In recent years, the field of microplastic (MP) research has evolved significantly; however, the lack of a standardized detection methodology has led to incomparability across studies. Addressing this gap, our current study innovates a reliable MP detection system that synergizes sample processing, machine learning, and optical photothermal infrared (O-PTIR) spectroscopy. This approach includes examining high-temperature filtration and alcohol treatment for reducing non-MP particles and utilizing a support vector machine (SVM) classifier focused on key wavenumbers that could discriminate between nylon MPs and non-nylon MPs (1077, 1541, 1635, 1711 cm−1 were selected based on the feature importance of SVM-Full wavenumber model) for enhanced MP identification. The SVM model built from key wavenumbers demonstrates a high accuracy rate of 91.33%. Results show that alcohol treatment is effective in minimizing non-MP particles, while filtration at 70 °C has limited impact. Additionally, this method was applied to assess MPs released from commercial nylon teabags, revealing an average release of 106 particles per teabag. This research integrates machine learning with O-PTIR spectroscopy, paving the way for potential standardization in MP detection methodologies and providing vital insights into their environmental and health implications. [ABSTRACT FROM AUTHOR]

Chemical image fusion refers to the combination of chemical images from different modalities for improved characterisation of a sample. Challenges associated with existing approaches include: difficulties with imaging the same sample area or having identical pixels across microscopic modalities, lack of prior knowledge of sample composition and lack of knowledge regarding correlation between modalities for a given sample. In addition, the multivariate structure of chemical images is often overlooked when fusion is carried out. We address these challenges by proposing a framework for multivariate chemical image fusion of vibrational spectroscopic imaging modalities, demonstrating the approach for image registration, fusion and resolution enhancement of chemical images obtained with IR and Raman microscopy. [ABSTRACT FROM AUTHOR]

This review focuses on the recent developments in vibrational spectroscopy and chemical imaging (i.e. Raman, Near Infrared, Mid Infrared) to characterize polymers in diverse forms, their behaviour and transient phenomenon. First, important polymeric properties and traditional methods of their characterization are outlined. Then relative advantages & disadvantages have been presented of different characterization methods are presented. This is followed by a detailed review of applications of chemical imaging and spectroscopic techniques in polymer characterization, including the limitations encountered. The article ends with a discussion on the future of chemical imaging with regards to polymer characterization. [ABSTRACT FROM AUTHOR]

Hyperspectral chemical imaging (HSI) is a broad term encompassing spatially resolved spectral data obtained through a variety of modalities (e.g. Raman scattering, Fourier transform infrared microscopy, fluorescence and near-infrared chemical imaging). It goes beyond the capabilities of conventional imaging and spectroscopy by obtaining spatially resolved spectra from objects at spatial resolutions varying from the level of single cells up to macroscopic objects (e.g. foods). In tandem with recent developments in instrumentation and sampling protocols, applications of HSI in microbiology have increased rapidly. This article gives a brief overview of the fundamentals of HSI and a comprehensive review of applications of HSI in microbiology over the past 10 years. Technical challenges and future perspectives for these techniques are also discussed. [ABSTRACT FROM AUTHOR]

Abstract: Identification and proper labelling of genetically modified organisms is required and increasingly demanded by legislation and consumers worldwide. In this study, the feasibility of three near infrared reflectance technologies (a chemical imaging unit, a commercial diode array instrument, and a light tube non-commercial instrument) were compared for discriminating Roundup Ready® and not genetically modified soybean seeds. Over 200 seeds of each class (Roundup Ready® and conventional) were used. Principal Component Analysis with Artificial Neural Networks (PCA–ANN) and Locally Weighted Principal Component Regression (LW-PCR) were used for creating the discrimination models. Discrimination accuracies when new tested seeds belonged to samples included in the training sets achieved accuracies over 90% of correctly classified seeds for LW-PCR models. The light tube performed the best, while the imaging unit showed the worse accuracies overall. Models validated with new seeds from samples not included in the training set had accuracies of 72–79%. [Copyright &y& Elsevier]

Abstract: Hyperspectral imaging (HSI) combines spectroscopy and imaging resulting in three dimensional multivariate data structures (‘hypercubes’). Each pixel in a hypercube contains a spectrum representing its light absorbing and scattering properties. This spectrum can be used to estimate chemical composition and/or physical properties of the spatial region represented by that pixel. One of the advantages of HSI is the large volume of data available in each hypercube with which to create calibration and training sets. This is also known as the curse of dimensionality, due to the resultant high computational load of high dimensional data. It is desirable to decrease the computational burden implied in hyperspectral imaging; this is especially relevant in the development of real time applications. This paper gives an overview of some pertinent issues for the handling of HSI data. Computational considerations involved in acquiring and managing HSI data are discussed and an overview of the multivariate analysis methods available for reducing the considerable data load encountered in HSI data is presented. [Copyright &y& Elsevier]

Abstract: The potential of hyperspectral imaging (HSI) in the visible-near infrared (445–945nm) wavelength range to discriminate between casing soil, enzymatic browning and undamaged tissue on mushroom (Agaricus bisporus) surfaces was investigated. A calibration set of 108 damage free mushrooms, grown under controlled conditions in a research station, were first tested as undamaged class (U) and then were divided into 2 groups of 54 samples. The first group was smeared with casing soil and designated as casing soil class (C) and the second group was subjected to vibrational damage resulting in enzymatic browning and designated as damaged class (D). Partial least squares discriminant analysis (PLS-DA) models were developed to classify mushroom tissue as one of the three classes investigated (U, C and D) using pixel spectra from each class. Prediction maps were obtained by applying the developed models to the hyperspectral images of candidate mushrooms. Percentages of pixels classified into each class were also calculated for the mushrooms studied in the calibration set. Results obtained showed that the developed models performed satisfactorily to discriminate between the 3 classes studied. Comparison of red–green–blue (RGB) and hyperspectral image analysis showed that HSI was better able to identify the regions containing casing soil. Model validation was performed using 3 different test sets of mushrooms obtained from a commercial producer. It was found that the developed PLS-DA models were satisfactorily capable of identifying undamaged regions, casing soil and enzymatic damaged areas on mushrooms from the validation sets. [Copyright &y& Elsevier]

The potential of hyperspectral imaging in the Vis-NIR range (400–1000 nm) for evaluating the quality of button mushrooms was compared to conventional RGB imaging using a digital colour camera. Regression models were built to correlate hyperspectral and RGB imaging data with measured Hunter L colour values. Prediction maps were generated from both hyperspectral and RGB imaging data to compare performance of the models at pixel level. Results presented showed that hyperspectral imaging is more suitable than conventional RGB imaging for evaluating mushroom quality. [ABSTRACT FROM AUTHOR]

Abstract: The shelf life of mushrooms packaged using different polymer top-films (PVC, PET with different levels of perforations) was investigated using hyperspectral imaging (HSI). Packaged mushrooms were stored at 4±0.2°C for 14days and weight loss, Hunter L, a, b values, maturity index and in-pack gas composition (% CO2 and O2) were also measured. The results obtained showed that the PET film perforated with small holes (1mm in diameter) was generally superior in terms of maintaining overall mushroom quality. Regression models were built to correlate HSI data with measured quality parameters. Prediction maps were generated from hyperspectral data to show the model performance at pixel level. Results presented in this work show hyperspectral imaging can be used to evaluate the effect of different packaging systems on mushroom quality and that perforated PET packaging film is a viable alternative to the conventional PVC packaging, facilitating an increase in shelf life from 10 to 14days. Industrial relevance: The present study demonstrates HSI can be used for rapid evaluation of mushroom quality facilitating non-destructive evaluation of the effect of packaging systems on mushroom shelf-life. In addition, this work suggests an effective packaging solution to extend shelf life of mushrooms during storage. The proposed solution potentially improves the packaging recyclability as the same polymer material (PET) is used for the tray and top film, compared to conventional mushroom packaging where PVC is used for the top film and polypropylene (PP) for the tray. [Copyright &y& Elsevier]

Abstract: In this study, the potential use of hyperspectral imaging (HSI) for early detection of freeze damage in white button mushrooms was investigated. Hyperspectral images of mushroom samples were obtained using a pushbroom line-scanning HSI instrument in the wavelength range of 400 to 1000nm. Reflectance spectra from each mushroom sample were obtained from various positions on the mushroom surface and pretreated using the Standard Normal Variate (SNV) transformation. Weight loss (WL) and Hunter L-value of mushroom samples were also measured over the storage period. At early stages of thawing, frozen samples were similar in appearance, WL and L-value to undamaged mushrooms, while substantial changes in WL and L-value were evident in freeze-damaged mushrooms after 24h thawing. A procedure based on Principal Components Analysis and Linear Discriminant Analysis was developed for classification of both selected spectra and whole mushrooms into undamaged and freeze-damaged groups. Applying this procedure to an independent test set, 100% of whole undamaged mushrooms were correctly classified and 97.9% of freeze-damaged samples were correctly classified. Using this method, freeze-damaged mushrooms could be classified with high accuracy (>95% correct classification) after only 45min thawing (at 23±2°C) at which time freeze–thaw damage was not visibly evident. [Copyright &y& Elsevier]

The objective of this study is to use a portable visible spectral imaging system (443–726 nm) to detect poultry thawed from frozen at the pixel level using multivariate analysis methods commonly used in machine learning (decision tree, logistic regression, linear discriminant analysis [ LDA ], k-nearest neighbors [ KNN ], support vector machines [ SVM ]). The selection of the most suitable method is based on the amount of data required to build an accurate model, computational speed, and the robustness of the model. The training set consists of pixel spectra from packages of chicken thighs without plastic lidding to evaluate the robustness of the models when implemented on the test set with and without plastic lidding. Data subsets were created by randomly selecting 1, 5, 10, 20, and 50% of the pixel spectra of each sample for both the training and test data sets. The subsets of pixel spectra and the full training set were used to train the machine learning algorithms to evaluate how the amount of data influences computational time. Logistic regression was found to be the best algorithm for detecting poultry thawed from frozen with and without plastic lidding film. Although logistic regression and SVM both performed with the same high accuracy and sensitivity for all training subset sizes, the computational time needed to implement SVM makes it the less suitable algorithm for detecting poultry thawed from frozen with and without plastic lidding film. [ABSTRACT FROM AUTHOR]

Moisture content, shrinkage, water activity (A w), color and texture of cooked chickpeas and soybeans during convective, microwave and combined microwave–convective drying were studied. Combined drying was significantly (P < 0.05) faster than either convective or microwave drying, and resulted in less shrinkage of the dehydrated product. Rapid burning occurred when samples were dried below a A w of 0.27 ± 0.07 for chickpeas (P < 0.05), and 0.13 ± 0.04 for soybeans (P < 0.05). Both chickpeas and soybeans displayed a transitional behavior in texture when dried to a A w below 0.40 ± 0.10 (P < 0.05) for chickpeas, and below 0.63 ± 0.15 (P < 0.05) for soybeans, when samples became brittle. Shelf stable dehydrated chickpea and soybean products with low water activity (A w = 0.35) and good visual quality could be obtained within 14 min of combination drying. [ABSTRACT FROM AUTHOR]

• Time series FTIR spectroscopy was used to study water vapor sorption in different polymeric films. • PVOH sorbed many water vapor molecules, leading to swelling and dissolving of polymer crystallites. • The amount of water sorbed was higher in PLLA and PVC than in PE or PTFE due to hydrophilic groups. • Band shift of difference spectra indicated stronger hydrogen-bonding network in the PTFE matrix. This work investigates the nature of the molecular interactions between water vapor and polymers using time series Fourier transform infrared (FTIR) spectroscopy. A detailed analysis of the frequency shifts and relative peak intensities led to the conclusion that polyvinyl alcohol (PVOH) sorbed a large quantity of water vapor molecules, resulting in swelling and dissolving of polymer crystallites. Difference spectra were calculated to investigate spectral changes occurring upon sorption by dividing the spectra of polymers during the sorption time series by the spectrum of the dry sample and subsequently subtracting the water vapor spectrum. Based on the absorbance area of the OH stretching vibration region (4000–3000 c m - 1 ) in difference spectra, the amount of water sorbed was significantly higher in poly-L-lactic acid (PLLA) and polyvinyl chloride (PVC) than in polyethylene (PE) and polytetrafluoroethylene (PTFE), increasing with the hydrophilicity of the surface. The OH stretching band of difference spectra shifted from 3499 c m - 1 for PVC, to 3416 c m - 1 for PE and finally to 3387 c m - 1 for PTFE, indicating a more strengthened hydrogen-bonding network in the PTFE matrix upon water vapor sorption. [ABSTRACT FROM AUTHOR]

The topic of food is broad and global, thereby representing an influential sector of the economy. Motivated by the advent of Industry 4.0, massive potential exists to implement cutting-edge technologies in the food industry. Recent years have seen a growing interest towards the applications of augmented/mixed reality (AR/MR) in the food sector. An extensive search of online journals focusing on Scopus was conducted using terms including 'augmented reality', 'mixed reality' and 'food' in the search fields of Title, Abstract, and Keywords. Full paper reading was implemented and ineligible articles (i.e., non-English-language, review articles, not peer-reviewed and without full paper) were removed. Our systematic search resulted in 111 eligible articles, eight of which related to MR technology. There is an overall increasing trend in the number of publications appearing annually since the first relevant publication in 2010. Analysing these publications demonstrates the multidisciplinary nature of this technology which is closely linked to machine learning, computer vision, the Internet of Things (IoT), and artificial intelligence. Our findings also revealed that AR/MR technology is mainly applied in the following areas: dietary assessment, food nutrition and traceability, food sensory science, retail food chain applications, food education and learning, and precision farming. Furthermore, we highlight the limitations and analytical challenges that hinder the application of AR/MR to food-related research, such as the lack of reliable wireless connection and the difficulty in recognizing food objects in a complex environment, while also describing future research needs and directions. • Applications of AR/MR for food is summarized from 111 eligible articles published since 2010. • Applications of AR/MR is closely linked to advanced technologies such as IoT, machine learning. • AR/MR is mainly applied for food safety, sensory, retail chain, education, and precision farming. • Limitations and challenges that hinder food applications of AR/MR are highlighted. • Future research needs and directions for AR/MR developments in food areas are presented. [ABSTRACT FROM AUTHOR]

Hyperspectral imaging (HSI) emerges as a non-destructive and rapid analytical tool for assessing food quality, safety, and authenticity. This work aims to investigate the potential of combining the spectral and spatial features of HSI data with the aid of deep learning approach for the pixel-wise classification of food products. We applied two strategies for extracting spatial-spectral features: (1) directly applying three-dimensional convolution neural network (3-D CNN) model; (2) first performing principal component analysis (PCA) and then developing 2-D CNN model from the first few PCs. These two methods were compared in terms of efficiency and accuracy, exemplified through two case studies, i.e., classification of four sweet products and differentiation between white stripe ("myocommata") and red muscle ("myotome") pixels on salmon fillets. Results showed that combining spectral-spatial features significantly enhanced the overall accuracy for sweet dataset, compared to partial least square discriminant analysis (PLSDA) and support vector machine (SVM). Results also demonstrated that spectral pre-processing techniques prior to CNN model development can enhance the classification performance. This work will open the door for more research in the area of practical applications in food industry. [ABSTRACT FROM AUTHOR]

Silver nanoparticles (AgNP) emerged as a promising reagent for cancer therapy with oxidative stress implicated in the toxicity. Meanwhile, studies reported cold atmospheric plasma (CAP) generation of reactive oxygen and nitrogen species has selectivity towards cancer cells. Gold nanoparticles display synergistic cytotoxicity when combined with CAP against cancer cells but there is a paucity of information using AgNP, prompting to investigate the combined effects of CAP using dielectric barrier discharge system (voltage of 75 kV, current is 62.5 mA, duty cycle of 7.5kVA and input frequency of 50–60Hz) and 10 nm PVA-coated AgNP using U373MG Glioblastoma Multiforme cells. Cytotoxicity in U373MG cells was >100-fold greater when treated with both CAP and PVA-AgNP compared with either therapy alone (IC 50 of 4.30 μg/mL with PVA-AgNP alone compared with 0.07 μg/mL after 25s CAP and 0.01 μg/mL 40s CAP). Combined cytotoxicity was ROS-dependent and was prevented using N-Acetyl Cysteine. A novel darkfield spectral imaging method investigated and quantified AgNP uptake in cells determining significantly enhanced uptake, aggregation and subcellular accumulation following CAP treatment, which was confirmed and quantified using atomic absorption spectroscopy. The results indicate that CAP decreases nanoparticle size, decreases surface charge distribution of AgNP and induces uptake, aggregation and enhanced cytotoxicity in vitro. • Combination of PVA-AgNP and CAP result in a synergistic effect of more than 100-fold in GBM cells. • Reactive species necessary for enhanced cytotoxicity. • CAP induced changes in PVA-AgNP size and surface charge distribution. • Novel darkfield spectral imaging determined enhanced uptake, aggregation and subcellular accumulation of PVA-AgNP by CAP. [ABSTRACT FROM AUTHOR]

This work proposes a new method to treat spatial and spectral information interactively. The method extracts spatial features, ie, variogram, gray‐level co‐occurrence matrix (GLCM), histograms of oriented gradients (HOG), and local binary pattern (LBP) features, from each wavelength image of hypercube and principal component analysis (PCA) is applied on this spatial feature matrix to identify wavelength‐dependent variation in spatial patterns. Resultant image is obtained by projecting the score values to the original data. Three datasets, including a synthetic hyperspectral image (Dataset 1), a set of real hyperspectral images of salmon fillets (Dataset 2), and remote‐sensing images (Dataset 3), were utilized to evaluate the performance of the proposed method. Results from Dataset 1 showed that the spatial‐spectral methods had the potential of reducing baseline offset noise. Dataset 2 revealed that spatial‐spectral methods can alleviate noisy pixels with strong signal and reduce shadow effects. In addition, substantial improvements were obtained in case of classification between white stripe and red muscle pixels by using the HOG‐based approach with correct classification rate (CCR) of 0.97 compared with the models directly built from raw and standard normal variate (SNV) preprocessed spectra (CCR = 0.94). Samson image of Dataset 3 suggested the flexibility and effectiveness of the proposed method by improving CCR of 0.96 using conventional PCA on SNV pretreated spectra to 0.98 using GLCM‐based approach on SNV preprocessed spectra. Overall, experimental results demonstrated that the spatial‐spectral methods can improve the results found by using the spectral information alone because of the spatial information provided. This work proposes a new method to treat spatial and spectral information interactively. The proposed method was tested on three datasets, including synthetic and real‐world hyperspectral images. Results demonstrated that the introduced spatial‐spectral method (a) has the potential to reduce baseline offset noise, (b) can alleviate noisy pixels with strong signal, and (c) can improve the classification performance. In summary, the proposed spatial‐spectral method enables to extract meaningful and useful information from hyperspectral images. [ABSTRACT FROM AUTHOR]

Molecular-dynamics simulations have been performed for full liquid water adsorbed onto two planar silicon surfaces, with varying hydrogen- and hydroxyl-termination (mimicking different extents of hydrophobicity and hydrophophilicity). It was found that there was water-density 'ordering' perpendicular to both surfaces – heavily dependent on the degree of hydrophobicity. The position and the width of the three solvation layers closest to the different surfaces depends, again, on the hydrophobicity of the surface. IR spectra of the first monolayer of adsorbed water indicate similarities to more confined-water dynamical behaviour, but without becoming ice-like. Moving away from the surface, the water behaviour converges on that of liquid water, albeit with some intermediate characteristics; this was seen for both hydro-phobic and –philic surfaces. [ABSTRACT FROM AUTHOR]

The present study investigated spatial heterogeneity in magnesium oxychloride cements within a model of a mould using hyperspectral chemical imaging (HCI). The ability to inspect cements within a mould allows for the assessment of material formation in real time in addition to factors affecting ultimate material formation. Both macro scale NIR HCI and micro scale pixel-wise Raman chemical mapping were employed to characterise the same specimens. NIR imaging is rapid, however spectra are often convoluted through the overlapping of overtone peaks, which can make interpretation difficult. Raman spectra are more easily interpretable, however Raman imaging can suffer from slower acquisition times, particularly when the signal to noise ratio is relatively poor and the spatial resolution is high. To overcome the limitations of both, Raman/NIR data fusion techniques were explored and implemented. Spectra collected using both modalities were co-registered and intra and inter-modality peak correlations were investigated while k-means cluster patterns were compared. In addition, partial least squares regression models, built using NIR spectra, predicted chemical-identifying Raman peaks with an R2 of up to >0.98. As macro scale imaging presented greater data collection speeds, chemical prediction maps were built using NIR HCIs. [ABSTRACT FROM AUTHOR]

Visible and near-infrared spectral imaging is a key non-destructive technique for rapid assessment of biophysical traits of plants. A major challenge with close-range spectral imaging of plants is spectral variation arising from illumination effects, which may mask the signals due to physiochemical differences. In the present work, we describe a new scatter correction technique called variable sorting for normalisation (VSN) and compare its efficiency with that of the commonly used standard normal variate (SNV) technique for the removal of unwanted illumination effects. Spectral images of potato plants were used for testing the correction. The results showed that the VSN outperformed SNV in removing illumination effects from the images of plants. The results show that the VSN approach to illumination correction can support high-throughput plant phenotyping where spectral imaging is used as a continuous monitoring tool. • Illumination correction in spectral imaging of plants. • Variable sorting for normalisation (VSN) is demonstrated. • After correction plants part segmentation was enhanced. • VSN proved to be an easy approach to removing illumination effects. [ABSTRACT FROM AUTHOR]

This paper describes the feasibility of using short-wave infrared (SWIR) spectrometry to classify concretes by their water-to-cement (w/c) ratios and predict their density. Concrete spectra of three w/c ratios (50%, 65%, 80%) were studied in the 1300–2200 nm range. A Partial Least Square Discriminant Analysis model was developed from the spectra of 36 samples, resulting in an 89% correct classification for the 18 validation samples, thereby demonstrating that SWIR spectrometry can detect differences in initial w/c ratios for hardened concretes. Additionally, differences in density and compressive strength as a function of the w/c ratio could be indirectly estimated through SWIR spectrometry. [ABSTRACT FROM AUTHOR]

Both near infrared hyperspectral imaging (NIR-HSI) and Raman hyperspectral imaging (R-HSI) were investigated in combination with multivariate analysis to assess binary mixtures of food powders. NIR-HSI and R-HSI data of corn flour (CF), icing sugar (IS) and binary mixtures of both in concentrations of 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80% and 90% CF ( w /w) were used to develop CF and IS concentration prediction models. The best model for IS concentration prediction in CF using NIR-HSI was developed using PLS-R based on 7 selected bands (RMSE CV 0.58%, RPD CV 55.4, RMSE P 1.00%, RPD P 48.6, R 2 1.000, 2LV), while the best model for IS concentration prediction in CF using R-HSI was developed using PLS-R on 11 selected bands (RMSE CV 1.75%, RPD CV 18.4, RMSE P 1.26%, RPD P 24.9, R 2 0.997, 4LV). Mixing quality of samples was assessed using the standard deviation of NIR-HSI prediction maps of CF and IS mixtures (50:50 ( w /w)) mixed for 0, 1, 3, 7, 10, 15, and 30 s. The standard deviation reduced from 23.4% prior to mixing to 3.2% after 30 s of mixing. The prediction maps developed provide spatial information which can be used to evaluate mixture quality during both the convective and diffusion phases of mixing. Both hyperspectral imaging techniques investigated were demonstrated to have high potential for assessment of binary mixtures of food powders in process analytical technology applications. [ABSTRACT FROM AUTHOR]

Abstract: Array detectors improve data collection speed in hyperspectral chemical imaging, yet are prone to striping noise patterns in the image plane, which is difficult to remove. This type of noise affects spectral features and disturbs visual impression of the scene. We found that this type of noise depends on the material composition and setting parameters, ie, pixel size, and it also varies in accordance with the signal intensity of the observed wavelength. To address this, we proposed a new correction method on the basis of the application of chemometric techniques in the image plane of each wavelength. To verify the effectiveness of this method, infrared transmission images of the 2″ × 2″ positive, 1951 USAF Hi‐Resolution Target and biomaterial samples were obtained with a 16‐element ( 8 × 2) pixel array detector. Point detector images of some samples were also acquired and used as reference images. The proposed correction method produced substantial improvements in the visual impression of intensity images. Principal components analysis was performed to inspect spectral changes after preprocessing, and the results suggested that the major spectral features were not altered while the stripes on intensity images were removed. Spectral profiles and principal components analysis loadings inspection confirmed the smoothing ability of this correction method. As traditional preprocessing techniques, standard normal variate and derivative transformation were not able to remove line artifacts, especially on the biomaterial images. Overall, the proposed method was effective for removing striping noise patterns from infrared images with a minimal alteration of the valuable hyperspectral image information. [ABSTRACT FROM AUTHOR]

Hydrogels are an important class of biomaterials that can absorb large quantities of water. In this study, changes in hydration of natural hydrogels (agar, chitosan, gelatin, starch, and blends of each with chitosan) during storage and rehydration were studied by using near-infrared hyperspectral imaging (NIR-HSI).Moisture content was calculated based on changes in sample weight during hydration. The NIR-HSI data were acquired by using a push-broom system operating in diffuse reflectance in the wavelength range 943 to 1650 nm. A novel synthesis method was developed to enable common preparation of each hydrogel. Mean spectra obtained from the hyperspectral images were analyzed, and predictive models for moisture content were developed by using partial least squares regression. Models were compared in predictive performance by using an independent validation set of data. The optimal model in predictive performance was a 1 latent variable partial least squares regression model developed on second derivative and mean centered pseudoabsorbance data in the wavelength range 943 to 1272 nm. This model was applied to pixel spectra from samples in the validation set to inspect spatial variations during dehydration and rehydration. Challenges associated with NIR-HSI of hydrogels with a large variation in moisture content are discussed. [ABSTRACT FROM AUTHOR]

Imaging spectroscopy integrates conventional imaging and spectroscopy to attain both spatial and spectral information from an object. Processing and analysis of a hypercube can be a hard task. Robust methods for hyperspectral data classification are required, insensitive to imaging deficiencies and high input dimension. Mushrooms have a thin and porous epidermal structure, and are sensitive to handling and transportation practices. Mechanical damage triggers a browning process within the tissue changing its metabolic state. The objective is to quantify different levels of physical perturbation on the mushroom pilei, using near-infrared spectral images and machine learning approaches. An ANN classifier is implemented, whose input is a small set of vectors containing representative information, and output is the set of categorical labels that correspond to different levels of mechanical vibration. For obtaining a salient dataset for classifying the images, the Harris corner detection algorithm is employed. The advantage of using interest points is to replace an exhaustive search over the entire image space by a computation over a concise set of highly informative points. A frame-based classification approach is proposed and shown to produce an increase in the classification accuracy, since feature vectors regarded as single instances may not always carry sufficient discriminant information. Comparisons with statistical features computed from wavelet coefficients showed that interest points are more suitable in assessing mechanical perturbation. Comparisons on a classifier level with support vector machines showed that ANNs perform better for the specific application, implying a connection between the classification method and the underlying learning problem. Overall, the frame-based classification scheme reduced the misclassification rate. This approach is suited for challenging classification problems where the degree of class separation is variable, i.e., assessment of mechanical damage in mushrooms. [ABSTRACT FROM AUTHOR]

Molecular-dynamics simulations for silicon, hydrogen- and hydroxyl-terminated silicon in contact with liquid water, at 220 and 300 K, display water-density 'ordering' along the laboratory z-axis, emphasising the hydrophobicity of the different systems and the position of this first adsorbed layer. Density of states (DOS) of the oxygen and proton velocity correlation functions (VACFs) and infrared (IR) spectra of the first monolayer of adsorbed water, calculated via Fourier transformation, indicate similarities to more confined, ice-like dynamical behaviour (redolent of ice). It was observed that good qualitative agreement is obtained between the DOS for this first layer in all systems. The DOS for the lower-frequency zone indicates that for the interface studied (i.e., the first layer near the surface), the water molecules try to organise in a similar form, and that this form is intermediate between liquid water and ice. For IR spectra, scrutiny of the position of the highest-intensity peaks for the stretching and bending bands indicate that such water molecules in the first solvating layer are organised in an intermediate fashion between ice and liquid water. [ABSTRACT FROM AUTHOR]

Classical molecular-dynamics simulations have been performed to examine the interplay between ubiquitin and its hydration-water sub-layers, chiefly from a vibrational-mode and IR viewpoint—where we analyse individual sub-layers characteristics. The vibrational Density of States (VDOS) revealed that the first solvation sub-shell indicates a confined character therein. For layers of increasing distance from the surface, the adoption of greater bulk-like spectral behaviour was evident, suggesting that vibrational harmonisation to bulk occurs within 6–7 Å of the surface. [ABSTRACT FROM AUTHOR]

Over the last two decades, research in hyperspectral imaging has been increasing and its use in horticulture is expected to be spreading in the coming years. The emerging techniques are currently gaining interest of the research community. However, there are still challenges to the applicability. In this review we demonstrate that hyperspectral imaging can be used as an effective tool for fruit, vegetables and mushrooms in assessing quality parameters related to well defined variables that can be analysed in the laboratory, as well as complex properties such as maturity, ripeness, detection of biotic defects, physiological disorders, mechanical damages, and sensory quality. Therefore, this paper starts by giving an overview of the quality concept of produce, measuring principles, theory and analysis of hyperspectral imaging systems. Then, emerging techniques to monitor and assess quality parameters, both pre- and postharvest, are described, as well as applications of these are reviewed and discussed. Afterwards, this review proceeds by illustrating the current and potential use of artificial intelligence and its subdomains, machine learning and deep learning, for hyperspectral imaging analysis in horticulture. Lastly, some challenges and considerations for future research are highlighted, including improvement of data availability, possible solutions for an improved integration of artificial intelligence and the transfer of knowledge from research parameters to parameters relevant for industrial stakeholders. • Hyperspectral imaging is an effective tool for in assessing quality parameters. • The most abundantly used wavelengths are 601–850 nm, used in over 50% of studies. • PLSR and SVM are most commonly used models for machine learning. • Deep learning in combination with hyperspectral data is unexplored in horticulture. • Improving availability, AI knowledge transfer and standardisation is recommended. [ABSTRACT FROM AUTHOR]

This work investigates non-contact reflectance spectral imaging techniques, i.e. microscopic Fourier transform infrared (FTIR) imaging, macroscopic visible-near infrared (VNIR), and shortwave infrared (SWIR) spectral imaging, for the identification of bacteria on stainless steel. Spectral images of two Gram-positive (GP) bacteria (Bacillus subtilis (BS) and Lactobacillus plantarum (LP)), and three Gram-negative (GN) bacteria (Escherichia coli (EC), Cronobacter sakazakii (CS), and Pseudomonas fluorescens (PF)), were collected from dried suspensions of bacterial cells dropped onto stainless steel surfaces. Through the use of multiple independent biological replicates for model validation and testing, FTIR reflectance spectral imaging was found to provide excellent GP/GN classification accuracy (> 96%), while the fused VNIR-SWIR data yielded classification accuracy exceeding 80% when applied to the independent test sets. However, classification within gram type was far less reliable, with lower accuracies for classification within the GP (< 75%) and GN (≤ 51%) species when calibration models were applied to the independent test sets, underlining the importance of independent model validation when dealing with samples of high biological variability. [ABSTRACT FROM AUTHOR]

This work investigates non-contact reflectance spectral imaging techniques, i.e. microscopic Fourier transform infrared (FTIR) imaging, macroscopic visible-near infrared (VNIR), and shortwave infrared (SWIR) spectral imaging, for the identification of bacteria on stainless steel. Spectral images of two Gram-positive (GP) bacteria (Bacillus subtilis (BS) and Lactobacillus plantarum (LP)), and three Gram-negative (GN) bacteria (Escherichia coli (EC), Cronobacter sakazakii (CS), and Pseudomonas fluorescens (PF)), were collected from dried suspensions of bacterial cells dropped onto stainless steel surfaces. Through the use of multiple independent biological replicates for model validation and testing, FTIR reflectance spectral imaging was found to provide excellent GP/GN classification accuracy (> 96%), while the fused VNIR-SWIR data yielded classification accuracy exceeding 80% when applied to the independent test sets. However, classification within gram type was far less reliable, with lower accuracies for classification within the GP (< 75%) and GN (≤ 51%) species when calibration models were applied to the independent test sets, underlining the importance of independent model validation when dealing with samples of high biological variability. [ABSTRACT FROM AUTHOR]

Near-infrared (NIR) chemical imaging is an emerging technique with the potential for the detection of contaminants in the environmental field. In this study the potential of NIR chemical imaging (NIR-CI) to predict concentrations of nutrients (total nitrogen, total phosphorus) and indicator microorganisms (Escherichia coli) in surface water was investigated. Chemical images of multiple samples were obtained simultaneously using a pushbroom imaging system operating in the 950–1650 nm wavelength range with spectral resolution of 7 nm. Using partial least squares regression models, the relationship between these pollutants and NIR spectral data extracted from the chemical images in samples of aqueous surface water and filtered residue from surface water was assessed. When calibration models were tested on an independent data set, it was found that models developed on filtered residue spectra outperformed those developed on aqueous samples. For samples of filtered residue, the performance of the calibrations achieved for total nitrogen was reasonable (R2 > 0.75); however, performance for total phosphorus andE. coliwas poor (R2 < 0.5). Lower concentrations of these parameters were detected in the surface water samples included in the study (<1 mg L−1and <20 colony-forming units per 100 mL, respectively), a likely reason for the poor performance. The results indicate that NIR-CI has the potential for screening samples in which the contaminant concentration exceeds 1 mg L−1. [ABSTRACT FROM PUBLISHER]

Abstract: The objective of this study was to evaluate a vat wall-mounted image capture system with a range of image processing techniques (threshold, first order and second order grey level statistics and fractal dimension) to monitor curd moisture content during syneresis with a range of temperature treatments. Milk was renneted using three temperature treatments (32°C throughout, a cooking step from 32 to 38°C and 38°C throughout). Prediction models were evaluated in term of fit to reference measurements on samples of curd at 10min intervals. The best fitting model was based on the threshold technique giving a standard error of prediction (SEP)=1.06g/100g and correlation coefficient (R)=0.90. These results demonstrated that the threshold image processing technique was the most useful in this study and showed adequate potential to monitor syneresis and predict curd moisture which influences the final texture of cheese. [Copyright &y& Elsevier]

The diagnosis of prostate cancer is challenging due to the heterogeneity of its presentations, leading to the over diagnosis and treatment of non-clinically important disease. Accurate diagnosis can directly benefit a patient's quality of life and prognosis. Towards addressing this issue, we present a learning model for the automatic identification of prostate cancer. While many prostate cancer studies have adopted Raman spectroscopy approaches, none have utilised the combination of Raman Chemical Imaging (RCI) and other imaging modalities. This study uses multimodal images formed from stained Digital Histopathology (DP) and unstained RCI. The approach was developed and tested on a set of 178 clinical samples from 32 patients, containing a range of non-cancerous, Gleason grade 3 (G3) and grade 4 (G4) tissue microarray samples. For each histological sample, there is a pathologist labelled DP–RCI image pair. The hypothesis tested was whether multimodal image models can outperform single modality baseline models in terms of diagnostic accuracy. Binary non-cancer/cancer models and the more challenging G3/G4 differentiation were investigated. Regarding G3/G4 classification, the multimodal approach achieved a sensitivity of 73.8% and specificity of 88.1% while the baseline DP model showed a sensitivity and specificity of 54.1% and 84.7% respectively. The multimodal approach demonstrated a statistically significant 12.7% AUC advantage over the baseline with a value of 85.8% compared to 73.1%, also outperforming models based solely on RCI and mean and median Raman spectra. Feature fusion of DP and RCI does not improve the more trivial task of tumour identification but does deliver an observed advantage in G3/G4 discrimination. Building on these promising findings, future work could include the acquisition of larger datasets for enhanced model generalization. [ABSTRACT FROM AUTHOR]

Time series spectral imaging facilitates a comprehensive understanding of the underlying dynamics of multi-component systems and processes. Most existing classification strategies focus exclusively on the spectral features and they tend to fail when spectra between classes closely resemble each other. This work proposes a hybrid approach of principal component analysis (PCA) and deep learning (i.e., long short-term memory (LSTM) model) for incorporating and utilizing the combined multi-temporal and spectral information from time series spectral imaging datasets. An example data, consisting of times series spectral images of casein-based biopolymers, was used to illustrate and evaluate the proposed hybrid approach. Compared to using partial least squares discriminant analysis (PLSDA), the proposed PCA-LSTM method applying the same spectral pretreatment achieved substantial improvement in the pixel-wise classification (i.e., accuracy increased from 59.97% of PLSDA to 85.73% of PCA-LSTM). When projecting the pixel-wise model to object-based classification, the PCA-LSTM approach produced an accuracy of 100%, correctly classifying the whole 21 film samples in the independent test set, while PLSDA only led to an accuracy of 80.95%. The proposed method is powerful and versatile in utilizing distinctive characteristics of time dependencies from multivariate time series dataset, which could be adapted to suit non-congruent images over time sequences as well as spectroscopic data. Image 1 • PCA-LSTM was proposed for classification of time series spectral images. • PCA-LSTM approach enables to incorporate joint multi-temporal and spectral features. • PCA-LSTM (CCR = 86%) outperformed PLSDA (CCR = 60%) for pixel-wise classification. • PCA-LSTM (CCR = 100%) outperformed PLSDA (CCR = 81%) for object-wise classification. [ABSTRACT FROM AUTHOR]

It is challenging to achieve consistent quantification of hydrogen‐bonded water species by fitting Gaussian bands on spectra collected from different instruments, especially when the spectral resolution is low. Therefore, this work aims to propose a near‐infrared (NIR) multivariate calibration model transfer method to obtain reliable, consistent, and reproducible results from different spectroscopic instruments. A composite method of Gaussian deconvolution, multivariate curve resolution‐alternating least squares (MCR‐ALS), and partial least squares (PLS) was developed to quantify five water species (labeled as C1, C2, C3, C4, and C5), respectively, assigned to non‐hydrogen‐bonded, singly hydrogen‐bonded, doubly and triply hydrogen‐bonded species, distorted ice‐like structures in the liquid phase, and fully hydrogen‐bonded species showing ice‐like structure. Primary instrument (iS50) was selected due to the highest spectral resolution and spectra (7500–5845 cm−1 with spectral resolution of 0.5 cm−1) were acquired from four types of salt solutions. The developed PLS model yielded a high prediction performance when applied on two independent prediction sets with R2P higher than 0.95 and RMSEP less than 0.009 for prediction of C1‐5. Model built with primary spectra was successfully transferred to spectra of secondary instrument (iN10), producing RMSEP less than 0.010. Transferred model applied on secondary instruments (NIRECO and MPA) showed that fully H‐bonded species in deionized water decreased as temperature increased, while non‐hydrogen‐bonded and singly hydrogen‐bonded demonstrated an opposite temperature dependence. Our results confirmed that it is promising to transfer the calibration model from a sophisticated spectrometer to a simple spectroscopic instrument in terms of quantification of hydrogen‐bonded water species. A composite method of Gaussian deconvolution, MCR‐ALS and PLS was proposed to quantify five hydrogen‐bonded water species. The developed model can be successfully transferred from a sophisticated spectrometer to a simple spectroscopic instrument. [ABSTRACT FROM AUTHOR]

Applying a calibration model onto hyperspectral (HS) images is of great interest because it produces images of chemical or physical properties. HS imaging is widely used in this way in food processing industries for monitoring product quality and process control. In this context, one of the main difficulties in the application of regression models to HS images is to evaluate the error of the obtained predictions, since in a proximal imaging set up, the size of the pixels is usually much smaller than the area required to obtain a wet chemical reference. Moreover, the selection of regression model parameters, such as the number of latent variables (LV) in a partial least squares (PLS) model, can modify the appearance of the prediction maps. The objective of this work is to propose an approach based on geostatistical indices to use spatial information of prediction maps for supporting the evaluation of regression models applied to HS images. This work stablishes a theoretical connection between linear regression model performance estimates and the spatial decomposition of variance in prediction maps, when the ground truth to estimate is spatially structured. This approach was tested in a simulated dataset and two real case studies. Geostatistical indices of the prediction maps were compared to model performance metrics for PLS models with increasing number of LV. The theoretical framework was proven by the results on the simulated dataset. In particular, the evolution of the nugget effect, C 0 , corresponded with the evolution of the random error of the model. Conversely, the error term of the model related with the slope of the model corresponded with the evolution of the structured variance observed in the prediction maps. On the real case studies, geostatistical indexes, extracted from the prediction maps, allowed to quantitatively evaluate the spatial structure of the estimations and complement the Root Mean Standard Error of Cross Validation (RMSECV) for the choice of optimal number of LV to consider in the model. The main advantage of this approach is that it does not require ground truth values. It could be used as a source of information for supporting the choice of optimum calibration parameters, such as the number of latent variables, or the choice of pre-treatments, complementing the traditional visual inspection of prediction maps with quantitative and objective metrics. Image 1 • How to evaluate model predictions from hyperspectral images? • We can use the spatial structure of prediction maps. • Spatially structured variance relates to the slope of the model. • Non spatially structured variance relates to the random error. • This method does not require to use ground truth values. [ABSTRACT FROM AUTHOR]

Highlights • A polarized-HSI system (400–1000 nm) was developed for agricultural application. • Superficial information contributes more in discriminating sunflower varieties. • Both surface and subsurface features are equally important in classification. • R ⊥ is most important in detecting diseases (PM and/or SLS) on sunflower leaves. • Best classification model for disease detection had CCR of 0.963 for prediction. Abstract This study aims to investigate the potential of an original polarized hyperspectral imaging (HSI) setup in the spectral domain of 400–1000 nm for sunflower leaves in real-world. Dataset 1 includes hypercubes of sunflower leaves in two varieties with different life growth stages, while Dataset 2 is comprised of healthy and contaminated sunflower leaves suffering from powdery mildew (PM) and/or septoria leaf spot (SLS). Cross polarised (R ⊥), parallel polarised (R | |) reflectance signals, R BS (R | | + R ⊥) and R SS (R | | - R ⊥) spectra were obtained and used to develop partial least squares-discriminant analysis (PLS-DA) models. Surface information played an important role in separating two varieties of leaves due to the fact that the best model performance was achieved by using R SS mean spectra, while both surface and subsurface were equally important in classifying leaves between two major growth stages because model of R BS mean spectra outperformed other models. The best classification model for disease detection was achieved by using pixel R ⊥ spectra with the correct classification rate (CCR) of 0.963 for both cross validation and prediction, meaning that subsurface spectral features were the most important to detect infected leaves. The resulting classification maps were also displayed to visualize the distribution of the infected regions on the leaf samples. The overall results obtained in this research showed that the developed polarized-HSI system coupled with multivariate analysis has considerable promise in agricultural real-world applications. [ABSTRACT FROM AUTHOR]

This study applies the single drop techniques to compare the efficacy of Raman, FT – NIR, and FT-MIR spectral imaging to quantify lactose concentration in dried whole milk on different metallic surfaces. Drying the samples avoids degradation problems such as water evaporation or oil degradation and scattering due to micelles. Spectral imaging techniques minimise sampling issues while also describing the sample spatial variation. The mean spectra of pre-processed images were used to build PLS regression models to predict lactose concentration. Raman, FT – NIR (5600–3730 cm−1), FT–MIR (3533–600 cm−1) models and the model obtained using the fusion of the three ranges were built independently and compared. This study confirms that is possible to quantify lactose rapidly using spectral imaging without adding standard references: the minimum RMSEP = 2.8 mg/mL (R2 = 0.98) was achieved with FT – MIR spectral imaging. [Display omitted] • Raman, FT-NIR, and FT-MIR spectral imaging comparison to quantify lactose in milk. • Spectral imaging approach minimises sampling issues. • Dehydration avoids degradation problems such as water evaporation or oil degradation. • FT – MIR spectral imaging quantifies lactose rapidly (RMSEP = 2.8 mg/mL, R2 = 0.98 mg/mL). [ABSTRACT FROM AUTHOR]

Deep learning (DL) is emerging as a new tool to model spectral data acquired in analytical experiments. Although applications are flourishing, there is also much interest currently observed in the scientific community on the use of DL for spectral data modelling. This paper provides a critical and comprehensive review of the major benefits, and potential pitfalls, of current DL tecnhiques used for spectral data modelling. Although this work focuses on DL for the modelling of near-infrared (NIR) spectral data in chemometric tasks, many of the findings can be expanded to cover other spectral techniques. Finally, empirical guidelines on the best practice for the use of DL for the modelling of spectral data are provided. • Status of deep learning for NIR data modelling is reviewed. • The hypes and benefits of deep learning for NIR data are highlighted. • Recommendations on fair DL practices are presented. [ABSTRACT FROM AUTHOR]

This work used portable spectral imaging to assess effects of ultraviolet (UV) (254 nm) decontamination and predict bacterial spoilage of vacuum packaged chicken breasts. Samples were scanned using spectral imaging, sampled for colour (L* - lightness, a* - redness, b* - yellowness), pH, total viable (TVC) and total Enterobacteriaceae counts (TEC) over 12 days of storage at 4 °C. UV treated samples had mean a* values 0.42 units significantly higher than untreated controls (p < 0.05). UV treatment affected the spectra of samples, allowing for discrimination between treatment types using Partial least squares discriminant analysis (PLS-DA) (CC% test = 97.5%). Values of TEC and TVC were significantly reduced by UV treatment and the effects remained significant over the storage time (p < 0.05), extending bacterial shelf life by 3 days. Partial least squares regression (PLSR) models were built from the measured spectra to predict TEC (R p 2 = 0.66, RMSEP = 0.63 log CFU/g) and TVC (R p 2 = 0.78, RMSEP = 0.87 log CFU/g). Interpretation of the spectra strongly suggested that UV treatment affected oxymyoglobin concentrations by accelerating myoglobin oxidation. However, despite differences in oxymyoglobin concentrations, it was still possible to build models using spectral data that could reasonably predict microbiological spoilage. • Portable spectral imaging evaluated the effect of UV light on vacuum packed chicken breasts. • UV treated samples were detected with a 97.5% correct classification rate. • TEC and TVC prediction was possible on an independent test set of samples. • Wavelengths pertinent to prediction are associated with oxymyoglobin concentration. [ABSTRACT FROM AUTHOR]