Near infrared spectroscopy (NIR), recognized as a rapid, non-invasive, easy-to-operate and cost-effective tool for substance analysis, has been widely used and intensively researched in the field of coffee bean analysis. From the perspective of multidimensional evaluation (physical assessment of green coffee, extrinsic assessment, descriptive assessment and affective assessment), this review systematically explores the current applications of near infrared spectroscopy in characterizing the attributes of Arabica coffee beans, and further investigates the identification of coffee bean adulteration and chemical properties through an exhaustive survey of more than 90 published, peer-reviewed articles. Building upon this foundation, this review analyzes and discusses the roles of chemometrics, spectral preprocessing methods, characteristic wavelength screening strategies and outlier removal methods. Through the combined application of these methods, the optimal NIR spectral inversion mechanisms corresponding to different coffee bean properties are revealed. This body of research not only highlights the substantial potential of near infrared spectroscopy but also pinpoints the pressing research gaps within the industrial process of coffee bean quality control. Currently, few studies have differentiated processing methods and elucidated the sweetness attributes, grading attributes, and flavor descriptors of coffee beans. Looking ahead, near infrared spectroscopy technology is anticipated to driving the establishment of a comprehensive and objective coffee bean value system.

Near infrared spectroscopy was used in discrimination of intact bovine teeth in terms of animal sex, diet, tooth type and place of origin. Discriminant Analysis (DA) models were developed and tested using a stratified random 70:30 data split to calibration and test sets. Of the discriminant techniques of Soft Independent Modelling of Class Analogy (SIMCA), Partial Least Squares – DA, Artificial Neural Networks (ANN) and Support Vector Machine (SVM), SVM and PLS-DA models performed best in most instances, with pretreatment choice impacting technique success. For test set prediction of animal sex, an accuracy of 95% was achieved using a PLS-DA model, and similar performance achieved using a SVM model. SVM, ANN and SIMCA models predicted three categories of tooth type (deciduous, permanent unerupted and erupted) with similar accuracies (89, 79 and 82%, respectively), however, permanent unerupted teeth were conflated with permanent erupted teeth. Prediction accuracy for ANN, SVM and PLS-DA models were also similar for discrimination of teeth from animals on a grain versus grass diet (PLS-DA 82%). There was considerable variation between models in prediction of geographic origin. Of the seven locations, two were discriminated with true positive rate around 70% using SVM and ANN models. Potential applications of this technology are discussed.

Hydroxyapatite is a major component of teeth and bones and is used commercially in metal sequestration. Near infrared imaging and spectroscopy has found increasing use in characterisation of these materials, particularly in context of dental conditions. The near infrared spectra of these materials are reviewed in terms of band assignments related to water in various states, P-OH and organic material, and in terms of light scattering. The effect of factors such as acid and heat on the NIR spectra of bones and teeth is also described. This review is intended to provide a resource for future researchers using NIR spectroscopy in characterisation of hydroxyapatite containing material.

The Part 1 prequel to this review evaluated the evolution of modelling techniques used in evaluation of fruit quality over the past three decades and noted a progression towards the use of artificial neural networks (ANNs) and convolutional neural networks (CNNs). In this review, Part 2, the use of CNNs for NIR fruit quality evaluation is explored, given the success of CNNs in various other fields, such as image, video, speech, and audio processing, and the availability of large (open source) datasets of fruit spectra and reference quality attribute, which is required for the training of CNN models. The review provides an overview of deep learning and the CNN architectures and techniques used in NIR spectroscopy for regression modelling, with advantages and disadvantages identified. Studies using CNN for NIR based fruit quality evaluation are then critically examined. Eight publications have presented on models using the same open-source mango dry matter calibration and test set, enabling inter-method comparisons. CNN models have been demonstrated to be accurate, precise and robust. Techniques of transfer learning for CNN models offer an alternative solution to model updating and calibration transfer methods applied in traditional chemometrics. The review has highlighted crucial areas that require resolution and exploration in this application through future research, including, (i) data requirements for training a CNN (ii) optimal spectral pre-processing for CNN (iii) CNN architecture and hyper-parameter selection and tuning for fruit quality evaluation (iv) CNN model interpretability and explainability. Future studies must conduct clearer comparison to partial least squares (PLS) regression and shallow ANNs to better assess the prospective benefit of using CNN, a more complex model. The potential for visualisation of spectra relevance to the CNN model using techniques such as GradCam, currently employed in visualising 2D-CNN models, remains to be explored.

Fourier-transform near infrared (FT-NIR) spectroscopy of ovarian tissue was used to predict maturity status of fish species with variable reproductive strategies collected at limited time periods of their spawning cycle. Reference data were derived from histologically prepared tissue samples from four species: Pacific cod ( Gadus macrocephalus), walleye pollock ( Gadus chalcogrammus) , Greenland turbot ( Reinhardtius hippoglossoides) , and northern rockfish ( Sebastes polyspinis). Each data set was classified into reproductively immature (non-spawning) and reproductively mature (spawning-capable) categories. Principal component analysis of spectral data showed separation between ovarian tissues of spawning-capable and non-spawning females. Multivariate classification using partial least squares discriminant analysis indicated good discrimination based on spawning status with high predictive accuracy. Greenland turbot and northern rockfish showed clear distinction between ovaries of spawning-capable and non-spawning females and a model validation with 100% and 96.6% classification accuracy, respectively. Pacific cod and walleye pollock had more complicated reproductive patterns at time of collection and classification rates were still 96.6% and 92.1%. This study demonstrated the potential application of FT-NIR spectroscopy to predict spawning status from ovarian tissue even for species with complicated spawning patterns and for collections outside of the preferred spawning period. Future work may include the use of this technology to classify distinct oocyte development stages.

Variable selection can improve the robustness and prediction accuracy of partial least squares (PLS) regression models and decrease the calculation time by selecting the optimal subset of variables in multivariate calibration. In this study, the performance of two variable selection methods for wavelength interval and individual wavelength coupled with partial least squares regression are investigated by employing the experimental data of asiaticoside (AS) and madecassoside (MS) contents in centella total glucosides (CTG) and a public dataset of corn. The studied variable selection methods include interval partial least squares regression ( iPLS), backward interval partial least squares ( biPLS), synergy interval partial least squares regression ( siPLS), competitive adaptive reweighted sampling (CARS), uninformative variable elimination (UVE) and variable importance in projection (VIP). The results show that the implementation of variable selection methods improved the performance of the model compared with full-spectrum modeling. All variable selection methods improved the prediction of AS or MS contents in CTG. When latent variables for PLS models are less than 10 in the practical application, the RPD value of AS models by iPLS method is 7.5, and the RPD value of MS models by biPLS method is 2.9. The results of wavelength interval selection are better than individual wavelength selection, especially for iPLS and biPLS. The same results were obtained with the public data for moisture in corn, and the RPD value of biPLS model of moisture is 1.6. Therefore, the wavelength interval selection methods, such as iPLS or biPLS, are appropriate for improving the PLS model’s accuracy and robustness to determine the target components’ contents in solid samples.

Non-invasive flesh firmness prediction using near infrared spectroscopy has been perfected and validated on three apple varieties. Three novel calibration models were developed following three year's of repeated large-scale sampling of stored commercial apple varieties ‘Gala’, ‘Red Jonaprince’ and ‘Jonagored’. The spectroscopic results were compared directly with those obtained using the invasive method. Increased accuracy of calibration models was achieved with the newly established data collection model. The results exhibited coefficient of determination for calibration, R2, and ratio of prediction to deviation (RPD) in excess of 0.91 and 2.3, respectively, thus enabling excellent prediction of flesh firmness via a non-invasive and fast spectroscopic approach. The highest R2 obtained was 0.94, RPD 2.6, root mean square error of calibration 5.87 N, and root mean square error of cross-validation (internal) 6.75 N for variety ‘Red Jonaprince’. Our complex long-term study provided excellent external validated calibration models and the approach can help developing calibration models for commercial sorting lines using near infrared spectroscopy.

This study uses a novel analysis methodology based on the Hierarchical Clustering Analysis (HCA) to determine the effectiveness of different preprocessing methods in minimizing undesired spectral variability in near infrared spectroscopy due to both the consecutive and repetitive acquisition of the spectrum and the sample temperature. Nine preprocessing methods and different combinations of them were evaluated in four case studies: reproducibility, repeatability, sample temperature, and combination of the before mentioned cases. Eighty-four spectra acquired on seven different hydrocarbon samples from catalytic conversion processes have been selected as the real case study to illustrate the potential of the mentioned methodology. The approach proposed allows a more detailed discriminatory analysis compared to the classical methods for comparing the between-class and the within-class variances, such as the Wilks’ lambda criterion, and hence constitutes a powerful tool to determine adequate spectral preprocessing strategies. This study also proves the potential of the discrimination analysis methodology as a general scheme to identify atypical behaviors either in the spectrum acquisition or in the measured samples.

Dried tangerine peel is a Chinese medicine with high medicinal value. The storage age is an important indicator of its medicinal value, so it is very significant to accurately identify the storage age of dried tangerine peel. Traditional physical and chemical analysis methods can be used to achieve this goal, but these methods are limited by their operability and convenience. Near infrared (NIR) spectroscopy and machine learning have excellent performance in the rapid detection of food and pharmaceutical samples. This study investigated the novel application of integrating a hand-held NIR spectrometer combined with machine learning to rapidly and accurately identify the storage age of Xinhui dried tangerine peel. Savitzky–Golay convolution smoothing, standard normal variate (SNV), first derivative, and second derivative pretreatments were employed to preprocess spectral data. Principal component analysis (PCA) was used to reduce the spectral data dimensions and obtain the characteristic spectral variables of each sample. Support vector machine (SVM) and k-nearest neighbor were applied to establish the qualitative discriminant models. The SNV-PCA-SVM model discriminant accuracy was 99.60% in the validation set and was 96.50% in the test set, showing excellent generalization performance. The results indicated that the method of using a hand-held NIR spectrometer combined with machine learning could be applied to rapidly identify the storage age of Xinhui dried tangerine peel. This is a promising and economical hand-held NIR spectroscopic method for assuring the dried tangerine peel age on-site.

After half a century of development, near infrared spectroscopy has now reached a relatively mature position. It is widely used in agriculture, food, petrochemical, and pharmaceutical fields and plays an increasingly important role in industrial and agricultural production and commercial trade. The development of Standards based on NIR spectroscopy represents the degree of recognition of the technology and applications. This article reviews domestic and global near infrared spectroscopy Standards in order to collate the myriad of existing standards in one listing.