•A MOS E-nose was applied to detect Chinese pecans at different storage times.•We proposed a novel voting method to classify different samples.•The accuracy rate of classification is improved to 96% by using voting method.•Six prediction models of fatty acids contents were built based on random forest.Metal oxide semiconductor (MOS) gas sensors have been widely used in the field of electronic nose (E-nose) for detecting the simple or complex volatile compounds. The electrical signals of MOS sensor array contain the abundant “fingerprint” information of samples. In this paper, an E-nose equipped with an array of MOS sensors was applied to detect Chinese pecans nondestructively for qualitative discrimination and quantitative prediction. Traditionally, most pattern recognition methods are based on single feature, which loses much useful information. To extract more information, a voting method, which was composed of principle component analysis (PCA) results based on 5 features (i.e., the 10th second values, the 75th second values, the area values, the maximum values and the minimum values of E-nose response curves), was proposed to classify pecan samples with different storage times. For quantitative prediction, six regression models were built on random forest (RF) algorithm to predict the contents of fatty acids in pecans. The analysis results showed that the voting method classified different pecan samples with 96% accuracy rate, and the regression models had satisfying prediction performance (R2 > 0.97 in calibration sets and R2 > 0.95 in validation sets). These results suggest that the voting method and RF algorithm would be promising analysis method for E-nose data.

The authors describe the application of two types of metallic foams modified with either graphene (GR) or carbon nanotubes (CNTs) as voltammetric electrodes in order to discriminate rice wines of different age and brand. Two types of bare metallic foams (bare Ni and Cu foam electrodes) were combined with GR or CNTs to give four types of modified metallic foams, referred to as GR/Ni, GR/Cu, CNT/Ni, and CNT/Cu foam electrodes. Cyclic voltammetry was applied to study the effects of GR and CNTs on the response of the electrodes. Multifrequency rectangle pulse voltammetry and multifrequency staircase pulse voltammetry were applied to generate potential waveforms, and chronoamperometric curves were recorded. Principal component analysis (PCA) allowed a classification of the rice wines, and characteristic regular distributions were identified in the PCA plots. Support vector machines (SVM) were found to perform better than partial least squares regression in predicting ages and brands of the rice wines in that all fit correlation coefficients were >0.9930. The SVM based leave-one-out cross-validation method proved to be the most powerful regression tool. The six types of foam electrodes perform very well in the classification and prediction of rice wines of different ages and brands.

•Egg storage time and yolk index were evaluated using electronic nose system.•Wavelet energy was extracted as feature signal of sensors for data analysis.•LDA and PNN methods performed successful classification on egg storage time.•Yolk index was predicted by SVM model with dimension reduction methods.Egg storage time and yolk index, two descriptors of egg freshness, were evaluated by an electronic nose combined with chemometric methods. To obtain more useful information from collected data, the wavelet energy was extracted as feature signal by the wavelet transform method for qualitative and quantitative analysis. For qualitative analysis, linear discriminant analysis (LDA) was applied to evaluate the feature signals, and the result indicated that these feature signals had good classification performance with the first two scores explaining 82.50% of total variance. Moreover, probabilistic neural network (PNN) was performed to classify eggs with different storage times, and 92.86% of samples in testing set were classified correctly. For quantitative analysis, back propagation neural networks (BPNN) and support vector machine (SVM) were applied to build prediction models of yolk index, indicating that SVM models (R2 = 0.9641 in training set and R2 = 0.8339 in testing set) were better than BPNN (R2 = 0.8629 in training set and R2 = 0.7863 in testing set). To further improve the performance of SVM models, independent component analysis (ICA) and local linear embedding (LLE) were used to reduce dimension of feature data, and the results showed that ICA-SVM model had satisfying prediction performance (R2 > 0.97).

Chinese pecans (Carya cathayensis) continuously deteriorate during storage because of their high fatty acid contents. In this study, an electronic nose (E-nose) was introduced to characterize Chinese pecans with different storage times. Chemometric methods (principal component analysis (PCA), partial least squares regression (PLSR), and back propagation neural networks (BPNNs)) were employed to analyze E-nose data. For qualitative analysis, PCA could visualize the discrimination between different pecans based on the E-nose data. For quantitative analysis, the results indicated that BPNN models performed better both in predicting storage times and fatty acid contents than the PLSR models. In addition, a multi-target BPNN regression model was built to simultaneously predict the contents of the six main fatty acids, and the results (R2 > 0.95 in calibration sets and R2 > 0.88 in validation sets) were satisfactory. This study provides a potentially viable method for determining the storage times and fatty acid profiles of nut products.

Correction for ‘Qualitative and quantitative analysis of fatty acid profiles of Chinese pecans (Carya cathayensis) during storage using an electronic nose combined with chemometric methods’ by Shui Jiang et al., RSC Adv., 2017, 7, 46461–46471.

•The HPP processed samples were assessed with E-nose.•Feature extraction methods (maximum value, area value and stable value) were applied.•Three linear dimension reduction methods were applied for feature selection.•ELM and Lib-SVM were applied for classification of HPP-processed mandarin juices.•ELM based on LPP showed better result than Lib-SVM based on other reduction methods.High Pressure Processing (HPP) is a high efficient method for food preservation. As an innovative inspection method, electronic nose (E-nose) was applied to assess the HPP mandarin juices in the headspace. In this work, mandarin juices were processed by HPP at range of 0–500 MPa and diagnosed by E-nose. To improve the efficiency of E-nose, locality preserving projections (LPP) was introduced to extract information of E-nose data and compared with principal component analysis (PCA) and linear discriminant analysis (LDA). Three data extractions (stable value, max value and area value), three data reductions (PCA, LDA and LPP) and two classification algorithms (support vector machine and extreme learning machine) were applied to improve the diagnostic accuracy of E-nose. This study shows that the performance of E-nose can be enhanced by appropriate data extraction and data reduction and also gives a hint of E-nose inspection for HPP fruit juice.Industrial relevanceHHP is a high efficient technology for food preservation, including volatile compounds and solid contents. E-nose, which detect the components changes in sample's headspace, has been widely applied for the food inspection as an innovative detection method. According to authors' best knowledge, researches about E-nose detection for HHP products are not very common. In this study, E-nose is applied to detect the HHP-processed mandarin juices in the headspace. Besides, three data extractions, three data reductions and two classification algorithms were applied to improve the accuracy of E-nose determination in two HHP-process cases. This paper describes the possible application of E-nose to discriminate HHP-processed juices at different pressures. E-nose combined with appropriate pattern recognition could be used as a rapid and high efficient method to discriminate HHP-processed fruit juices.

•Electronic nose was employed to detect tea plants with pest damage.•Difference in VOCs of tea plants with different treatments was proved using GC-MS.•A new feature extraction method was proposed and its strength was proved.•The results showed feasibility of electronic nose in pest damage detection.Damage of tea plant causes a lot of loss in tea production, but there is not an appropriate method to detect tea plants with pest damage. In this work, electronic nose (E-nose) and Gas Chromatography-Mass Spectrometer (GC-MS), as an auxiliary technique, were employed to detect tea plants with pest damage in two aspects, including tea plants with different invasive severities and with different invasive times, for giving a comprehensive results. A new feature extraction method based on a piecewise function was proposed and its performance was compared with those of the other three commonly employed models-polynomial functions, exponential functions, and Gaussian functions. Feature selection based on principal component analysis (PCA) and multi-layered perceptron (MLP) were employed for further feature reduction and classification, respectively. The results showed that feature extraction based on piecewise function was the best. The combination of feature extraction based on piecewise function, feature selection based on PCA and MLP was the best method and good enough for the classification in tea plants damage area. The results proved that E-nose was able to detect tea plants either with different invasive severities or different invasive times.Download high-res image (184KB)Download full-size image

•E-nose was proposed to predict the contents of benzoic acid and chitosan in juice.•Random forest (RF) and extreme learning machine (ELM) were used to process signals.•Support vector machine (SVM) and partial least squares regression (PLSR) were applied to treat signals.•Regression models based on RF and ELM showed higher prediction accuracy than SVM and PLSR.Food additives are added to products to enhance their taste, and preserve flavor or appearance. While their use should be restricted to achieve a technological benefit, the contents of food additives should be also strictly controlled. In this study, E-nose was applied as an alternative to traditional monitoring technologies for determining two food additives, namely benzoic acid and chitosan. For quantitative monitoring, support vector machine (SVM), random forest (RF), extreme learning machine (ELM) and partial least squares regression (PLSR) were applied to establish regression models between E-nose signals and the amount of food additives in fruit juices. The monitoring models based on ELM and RF reached higher correlation coefficients (R2s) and lower root mean square errors (RMSEs) than models based on PLSR and SVM. This work indicates that E-nose combined with RF or ELM can be a cost-effective, easy-to-build and rapid detection system for food additive monitoring.Download high-res image (123KB)Download full-size image

A taste sensing system was used to determine the marked ages and flavours of rice wines. The taste sensing system consisted of five different artificial lipid–polymer membrane electrodes that were highly sensitive to the five basic tastes (umami, astringency, bitterness, sourness and saltiness). This taste sensing system can be used to show the quality and intensity of tastes in samples and detects these tastes in a manner similar to that of the human gustatory system. Three types of rice wine with the same flavour, but different marked ages, and three types of rice wine with same marked age, but different flavours were analysed. A “taste map” analysis was performed to determine taste intensity based on which the tastes could be quantitatively analysed without chemometric methods. The differences in flavour among the rice wines were clearly shown. The responses (including the change in membrane potential caused by adsorption) obtained by the electrodes were analysed using principal components analysis and discriminant function analysis for classification and partial least-squares regression and a support vector machine for forecasting. Discriminant function analysis performed better than principal components analysis in classifying the rice wines with different marked ages and had similar results to principal components analysis in classifying rice wines with different flavours. The support vector machine based on the leave-one-out cross-validation was more stable than partial least-squares regression and the support vector machine based on the ten-fold cross-validation in predicting the marked ages and flavours of different types of rice wine; the prediction correlations were R2 = 0.9568 and R2 = 9620, respectively.

•We developed a sensitive PPy/TiO2 based gas sensor coated on QCM substrate.•We deposited the ultrathin film via simple layer-by-layer self-assembly approach.•Its outperformance was strongly depended on the thickness of film.•The obtained gas sensor was sensitive to designed toxic gases at low concentration.•The gas sensor could be applied in evaluating shelf-life of typical foodstuffs.An extra sensitive quartz crystal microbalance (QCM) gas sensor coated with thin PPy/TiO2 nanocomposite film was fabricated by using layer by layer self-assembly (SA) technology. The synthetic procedure and the resultant nanocomposites were characterized by using X-ray diffraction (XRD), transmission electron microscopy (TEM), Fourier transform infrared spectroscopy (FTIR) and field-emission scanning electron microscopy (FE-SEM). It was found that an ultra-sensitive PPy/TiO2 nanocomposite film with very thin layer can be successfully obtained by. It was also found that the number of deposited layers strongly impacted on sensor response with ten bilayers showing best sensor performance. The obtained gas sensor coating with PPy/TiO2 sensitive film was found to exhibit a better performance with respect to sensor responses, which is based on frequency data. The resultant sensor represented high sensitivity toward 10 ppm of different targeted gases with evident frequency shift, fast response and recovery time. Long-term stability and excellent reversibility were also observed. In real-time application, a designed measurement set-up based on PPy/TiO2 based sensor showed a good ability on shelf-life evaluation of foodstuffs (mango, egg and fish). The resulting QCM based gas sensor coated with PPy/TiO2 nanocomposite via Layer by Layer self-assembly presented a promising capability to detect trace irritant gases and food quality evaluation.

•Shelf lives of peanuts with and without pods were monitored by electronic nose.•Acid and peroxide values of peanut kernels were tested during storage time.•Three methods were employed to extract feature data from e-nose responses: maximum values, area values, and 70th s values.•Peanuts, with and without pods, were classified similarly in principal component analysis.•E-nose responses of unshelled peanuts had good relevance with acid and peroxide values.In this study, the changes in the quality of unshelled peanuts and peanut kernels during storage were analyzed using an electronic nose (e-nose). The physicochemical indexes (acid and peroxide values) of peanut kernels were tested by traditional method as a reference. The storage time of peanut kernels increases from left to right in the cluster analysis plot based on the physicochemical indexes. The “maximum values”, “area values”, and “70th s values” methods were applied to extract the feature data from the e-nose responses. Principal component analysis (PCA) results indicated that the “70th s values” method produced the most accurate results, furthermore, unshelled peanut and peanut kernel samples presented similar characteristics in the PCA plots; the partial least squares regression (PLSR) results showed that the features of unshelled peanuts and peanut kernels are highly correlated with acid and peroxide values, respectively.

•Aroma profiles of ginseng at different ages were studied by using E-nose and GC–MS.•The correlation between sensor response and identified compounds was investigated.•Closely related terpenes were selected using PLS loading and score analysis.•Content of most aroma contributed by terpenes can be predicted by gas sensors.•E-nose was performed better than GC–MS in predicting ginseng ages.Aroma profiles of ginseng samples at different ages were investigated using electronic nose (E-nose) and GC–MS techniques combined with chemometrics analysis. The bioactive ginsenoside and volatile oil content increased with age. E-nose performed well in the qualitative analyses. Both Principal Component Analysis (PCA) and Discriminant Functions Analysis (DFA) performed well when used to analyze ginseng samples, with the first two principal components (PCs) explaining 85.51% and the first two factors explaining 95.51% of the variations. Hierarchical Cluster Analysis (HCA) successfully clustered the different types of ginsengs into four groups. A total of 91 volatile constituents were identified. 50 of them were calculated and compared using GC–MS. The main fragrance ingredients were terpenes and alcohols, followed by aromatics and ester. The changes in terpenes, alcohols, aromatics, esters, and acids during the growth year once again confirmed the dominant role of terpenes. The Partial Least Squares (PLS) loading plot of gas sensors and aroma ingredients indicated that particular sensors were closely related to terpenes. The scores plot indicated that terpenes and its corresponding sensors contributed the most in grouping. As regards to quantitative analyze, 7 constituent of terpenes could be accurately explained and predicted by using gas sensors in PLS models. In predicting ginseng age using Back Propagation–Artificial Neural Networks (BP-ANN), E-nose data was found to predict more accurately than GC–MS data. E-nose measurement may be a potential method for determining ginseng age. The combination of GC–MS can help explain the hidden correlation between sensors and fragrance ingredients from two different viewpoints.

•We applied a semi-supervised approach—Cluster-then-Label—for e-nose data.•Spectral clustering was employed to improve the semi-supervised approach.•Semi-supervised approach was compared with supervised SVM, LDA, QDA and BPNN.•Semi-supervised approach outperformed supervised classifiers for three e-nose datasets.•The semi-supervised classifier requires only a few labeled e-nose data for training.Supervised classification, which is a fundamental classification approach for e-nose data, requires sufficient labeled data for training. However, sufficient labeled data requires extensive money, materials, energy and time. In this paper, a semi-supervised approach—Cluster-then-Label—that simultaneously uses labeled and unlabeled data to build a better classifier with fewer training data was introduced to deal with e-nose data for the first time. A novel clustering algorithm—spectral clustering—was also introduced to improve this semi-supervised approach. Three experiments—discriminating storage shelf life (SL), identifying pretreatments and authenticating juices, respectively—were conducted on cherry tomato juices using a PEN 2 e-nose, generating three datasets of different data structures. For each dataset, only 20% of data were selected for training. Classifications of the datasets by this semi-supervised approach and four supervised approaches (linear discriminant analysis (LDA), quadratic discriminant analysis, multi-class support vector machine and back propagation neural network) were compared. The results indicate that this spectral clustering based semi-supervised approach outperforms the supervised approaches in all cases. By using this semi-supervised approach, it is now possible to build reliable classifiers with only a few labeled data. It is also worth mentioning that this new approach takes no remarkable superiority over LDA. Thus, our next plan is to use more e-nose datasets for test.

Fruits freshness is relatively easy to authenticate from their morphological characteristics while the act of processing fruits into juices makes it difficult to track/identify their freshness. Eight datasets, extracted from an e-nose and an e-tongue, and six sensor fusion approaches using both instruments, were applied to detect 100 % juices squeezed from cherry tomatoes with different post-harvest storage times (ST). Discrimination of the juices was mainly performed by canonical discriminant analysis (CDA) and library support vector machines (Lib-SVM). Tracking and prediction of physicochemical qualities (pH, soluble solids content (SSC), Vitamin C (VC), and firmness) of the fruit were performed using principle components regression (PCR). All eight datasets presented good classification results with classifiers trained by e-tongue dataset and fusion dataset 2 (stepwise selection) presented the best classification performances. Though quality regression models trained by either e-nose or e-tongue dataset were not robustness enough, sensor fusion approaches make it possible to build more robust prediction models that can correctly predict quality indices for a totally new juice sample. This study indicates the potential for tracking quality/freshness of fruit squeezed for juice consumption using the e-nose and e-tongue, and that sensor fusion approach would be better than individual utilization only if proper fusion approaches are used.

An electronic nose (E-nose) and an electronic tongue (E-tongue) have been used to characterize five types of strawberry juices based on processing approaches (i.e., microwave pasteurization, steam blanching, high temperature short time pasteurization, frozen–thawed, and freshly squeezed). Juice quality parameters (vitamin C, pH, total soluble solid, total acid, and sugar/acid ratio) were detected by traditional measuring methods. Multivariate statistical methods (linear discriminant analysis (LDA) and partial least squares regression (PLSR)) and neural networks (Random Forest (RF) and Support Vector Machines) were employed to qualitative classification and quantitative regression. E-tongue system reached higher accuracy rates than E-nose did, and the simultaneous utilization did have an advantage in LDA classification and PLSR regression. According to cross-validation, RF has shown outstanding and indisputable performances in the qualitative and quantitative analysis. This work indicates that the simultaneous utilization of E-nose and E-tongue can discriminate processed fruit juices and predict quality parameters successfully for the beverage industry.

The freshness of fruit is relatively easy to authenticate by its morphological characteristics, while processing fruits into juices makes the freshness difficult to identify. In this paper, cherry tomatoes at different storage temperatures (4 and 25 °C) and shelf lives (SLs, 16 days at 4 °C and 8 days at 25 °C) were squeezed for use in 100% juices. Quality indices (SL, pH, soluble solids content (SSC), vitamin C (VC) concentration and firmness) of these cherry tomatoes were determined through analysing the juices using two sensor systems – an e-nose and an e-tongue. Support vector regression (SVR) was applied to predict the quality indices. The prediction performances based on a one sensor system, as well as a combination of two systems, were compared. The results showed that the e-tongue, which presents a similar prediction performance to the combination system, presents a better prediction performance (with higher squared correlation coefficients (R2) and a lower standard error of prediction (SEP)) than the e-nose. For tomatoes stored at 4 °C, the prediction parameters (R2, SEP) based on the e-tongue data for the SL, pH, SSC, VC concentration and firmness are (0.998, 0.295 d), (0.971, 0.022), (0.906, 0.075 °Brix), (0.978, 1.005 mg per 100 g) and (0.906, 0.292 N), respectively. For tomatoes stored at 25 °C, the prediction parameters (R2, SEP) based on the e-tongue data for the SL, pH, SSC, VC concentration and firmness are (0.997, 0.193 d), (0.934, 0.017), (0.957, 0.075 °Brix), (0.902, 0.897 mg per 100 g) and (0.908, 0.593 N), respectively. These results prove that it is possible to measure the freshness of fruits that are squeezed for juice consumption using sensor systems, and that the combination of sensor systems is not always better than using a one sensor system.

•We applied a novel clustering method – spectral clustering – for e-nose.•We conducted three e-nose experiments, generating three independent datasets.•We applied three cluster validation criteria to quantify clustering results.•Spectral clustering outperformed K-clustering and hierarchical clustering.Various clustering algorithms have been developed since conventional hierarchical cluster analysis (HCA) and partitioning clustering algorithms have their own limitations and scopes of applications. However, in the area of e-nose where clustering is applied, the conventional algorithms (mostly HCA) still play a dominant role. In addition, comparison among different clustering methods or validation of clustering results was seldom mentioned. In this paper, we present a state-of-the-art clustering method – spectral clustering – and compare it with six conventional clustering methods: K-clustering (ISODATA, FCM and k-means) and HCA (single linkage, complete linkage and Ward's). Three external validation criteria – mutual information criteria (MI), precision and rand index (RI) – were used to evaluate clustering performances on three independent e-nose datasets. The spectral clustering outperforms with statistical significance (alpha = 0.05) the performance of other methods, and the single linkage presents the worst (unacceptable) clustering result. In addition, the proposed approach – cluster validation criteria in combination with majority voting – in a way makes clustering a semi-supervised classification technique. Using this approach it is possible to compare clustering based semi-supervised methods with classification methods to find which method is better for discrimination of a certain e-nose dataset.

A voltammetric electronic tongue (VE-tongue) was self-developed and applied to monitor the quality and storage time of unsealed pasteurized milk. The VE-tongue comprised four working electrodes: gold, silver, platinum, and palladium electrode. Two potential waveforms: Multi-frequency rectangle pulse voltammetry (MRPV) and multi-frequency staircase pulse voltammetry (MSPV) were applied to working electrodes in the study, and both of MRPV and MSPV consisted of three frequency segments: 1 Hz, 10 Hz, and 100 Hz. The total areas under the corresponding curves obtained by VE-tongue in the three frequencies were applied as characteristic data, which were evaluated by the principal component analysis (PCA) and cluster analysis (CA). The results of PCA and CA indicate that the milk samples of different storage time could be successfully classified by the VE-tongue based on MRPV and MSPV, respectively. Combining the areas obtained by the VE-tongue based on MRPV and MSPV, the classification results of PCA and CA were improved evidently. The total bacterial count, acidity and viscosity of the milk samples were also measured during the storage, and those physicochemical characteristics showed regular configuration in PCA and CA plots. Furthermore, the total bacterial count and viscosity properties were predicted by partial least squares regression (PLSR) and least squares-support vector machines (LS-SVM), and the combination of the areas obtained by the VE-tongue based on the MRPV and MSPV were applied as the input data of PLSR and LS-SVM. Both the prediction techniques performed well in predicting viscosity and total bacterial count, and the prediction results of LS-SVM were better than that of PLSR. Those results demonstrate that the VE-tongue could be applied to monitor the quality storage time of unsealed pasteurized milk.

This paper describes the application of the electronic tongue (E-tongue) in the Discrimination of different white chrysanthemum. Three grade samples, two brands of samples and the samples adulterated were measured by the E-tongue. It was found the samples with different grades or brands could be clearly discriminated and the samples adulterated were separated from the authentic samples using PCA. The results of DFA and BPNN showed the total predicted accuracy of three grades samples were 86.7% and 93.3%. A strong positive correlation was observed between the sensory score and the predicted score (correlation coefficient is 0.9768) using PLS, and the samples were correctly classified. These results suggest the E-tongue may be useful for quality control of white chrysanthemum.

In this study, responses of a sensor array were employed to establish a quality index model able to describe the different picking date of peaches. The principal component regression (PCR) and partial least-squares regressions (PLS) model represent very good ability in describing the quality indices of the selected three sets of peaches in calibration and prediction. The results showed that the PLS model represents a good ability in predicting quality index, with high correlation coefficients (R = 0.86 for penetrating force [CF]; R = 0.83 for sugar content [SC]; R = 0.83 for pH) and relatively low standard error of prediction (SEP; 8.77 N, 0.299 °Brix, and 0.2 for CF, SC, and pH, respectively). The PCR model had high correlation coefficients (R = 0.84, 0.82, 0.78 for CF, SC, and pH, respectively) between predicted and measured values and a relatively low SEP (7.33 N, 0.44 °Brix, 0.21 for CF, SC, and pH, respectively) for prediction. These results prove that the electronic noses have the potential to assess fruit quality indices.

An electronic nose (e-nose) instrument combined with chemometrics was used to predict the physical–chemical indexes (sensory scores, total volatile basic nitrogen (TVBN) and microbial population) for beef. The e-nose data generated were analyzed by chemometrics methods and pattern recognition. Mahalanobis Distance (MD) analysis, Principle Component Analysis (PCA) and Linear Discriminant Analysis (LDA) confirmed the difference in volatile profiles of beef samples of 7 different storage times (ST). The Back Propagation Neural Network (BPNN) and Generalized Regression Neural Network (GRNN) were used to build prediction models for ST, TVBN content, microbial population and sensory scores. The result of GRNN was better than that of BPNN, and the standard error (SE) of GRNN prediction model for ST, TVBN, microbial population, sensory scores were 1.36 days, 4.64 × 10−2 mg g−1, 1.612 × 106 cfu g−1 and 1.31 respectively. This research indicates that it is of feasibility to use e-nose to predict multiple freshness indexes for beef.

A voltammetric electronic tongue (VE-tongue) based on multifrequency large amplitude pulse voltammetry (MLAPV) was developed to classify monofloral honeys of seven kinds of floral origins. The VE-tongue was composed of six working electrodes (gold, silver, platinum, palladium, tungsten, and titanium) in a standard three-electrode configuration. The applied waveform of MLAPV was composed of four individual frequencies: 1 Hz, 10 Hz, 100 Hz, and 1000 Hz. Two eigenvalues (the maximum value and the minimum value) of each cycle were extracted for building the first database (FDB); four eigenvalues (the maximum value, the minimum value, and two inflexion values) were exacted for building the second database (SDB). The two databases were analyzed by three-pattern recognition techniques: principal component analysis (PCA), discriminant function analysis (DFA) and cluster analysis (CA), respectively. It was possible to discriminate the seven kinds of honeys of different floral origins completely based on FDB and SDB by PCA, DFA and CA, and FDB was certificated as an efficient database by contrasting with the SDB. Moreover, the effective working electrodes and frequencies were picked out as the best experimental project for the further study.Highlights► We self-developed a voltammetric electronic tongue based on new sensors array. ► We advanced a new method to extract eigenvalues from signals obtained by VE-tongue. ► We first detected the monofloral honeys of different floral origins using VE-tongue.

A voltammetric electronic tongue (VE-tongue) was developed to detect antibiotic residues in bovine milk. Six antibiotics (Chloramphenicol, Erythromycin, Kanamycin sulfate, Neomycin sulfate, Streptomycin sulfate and Tetracycline HCl) spiked at four different concentration levels (0.5, 1, 1.5 and 2 maximum residue limits (MRLs)) were classified based on VE-tongue by two pattern recognition methods: principal component analysis (PCA) and discriminant function analysis (DFA). The VE-tongue was composed of five working electrodes (gold, silver, platinum, palladium, and titanium) positioned in a standard three-electrode configuration. The Multi-frequency large amplitude pulse voltammetry (MLAPV) which consisted of four segments (1 Hz, 10 Hz, 100 Hz and 1000 Hz) was applied as potential waveform. The six antibiotics at the MRLs could not be separated from bovine milk completely by PCA, but all the samples were demarcated clearly by DFA. Three regression models: Principal Component Regression Analysis (PCR), Partial Least Squares Regression (PLSR), and Least Squares-Support Vector Machines (LS-SVM) were used for concentrations of antibiotics prediction. All the regression models performed well, and PCR had the most stable results.

A voltammetric electronic tongue (VE-tongue) was developed to discriminate the difference between Chinese rice wines in this research. Three types of Chinese rice wine with different marked ages (1, 3, and 5 years) were classified by the VE-tongue by principal component analysis (PCA) and cluster analysis (CA). The VE-tongue consisted of six working electrodes (gold, silver, platinum, palladium, tungsten, and titanium) in a standard three-electrode configuration. The multi-frequency large amplitude pulse voltammetry (MLAPV), which consisted of four segments of 1 Hz, 10 Hz, 100 Hz, and 1000 Hz, was applied as the potential waveform. The three types of Chinese rice wine could be classified accurately by PCA and CA, and some interesting regularity is shown in the score plots with the help of PCA. Two regression models, partial least squares (PLS) and back-error propagation-artificial neural network (BP–ANN), were used for wine age prediction. The regression results showed that the marked ages of the three types of Chinese rice wine were successfully predicted using PLS and BP–ANN.Highlights► We self-developed a VE-tongue based on six metallic electrodes. ► We used Multi-frequency large amplitude pulse voltammetry as scanning waveforms. ► We first analyzed Chinese rice wine using the six metallic electrodes.

The impact force response of a peach impacting on a metal flat-surface was considered as nondestructive determination of firmness. The objectives were to analyze the effect of firmness, drop height, fruit mass, and impact orientation on the impact force parameters, and to establish a relationship between the impact force parameter and firmness. The effect of fruit firmness, drop height and fruit mass on the impact force parameters (coefficient of restitution, percentage of energy absorbed, and coefficient of force-time) was evaluated. The study found that the coefficient of restitution, percentage of energy absorbed, and force-time impact coefficient were significantly affected by fruit ripeness, but not affected by drop height, impact position (fruit cheek), and mass. The percentage of absorbed energy increased with ripeness, while the force-time impact coefficient and coefficient of restitution decreased with ripeness. Relationships were obtained between the three impact characteristic parameters (force-time impact coefficient, coefficient of restitution, and percentage of energy absorbed) and peach firmness using a polynomial model (R2=0.932), S model (R2=0.910), and exponential model (R2=0.941), respectively.

The objective of this study was to investigate the predictability of an electronic nose for fruit quality indices. Responses signal of sensor array in electronic nose were employed to establish quality indices model for “xueqing” pear. The relationships were established between signal of electronic nose and the quality indices of fruit (firmness, soluble solids content (SSC) and pH) by multiple linear regressions (MLR) and artificial neural network (ANN). The prediction models for firmness and soluble solids content indicated a good prediction performance. The SSC model by ANN had a standard error of prediction (SEP) of 0.41 and correlation coefficient 0.93 between predicted and measured values, the model by ANN for the penetrating force (CF) had a 3.12 SEP and 0.94 coefficient, respectively. The results imply that it is possible to predict “xueqing” pear quality characteristics from signal of E-nose.

In this paper, responses of a gas sensor array were employed to establish a quality indices model evaluating the peach quality indices. The relationship between sensor signals and the firmness, the content of sugar (CS) and acidity of “Dabai” peach were developed using multiple linear regressions with stepwise procedure, quadratic polynomial step regression (QPST) and back-propagation network. The results showed that the multiple linear regression model represented good ability in predicting of quality indices, with high correlation coefficients (R2 = 0.87 for penetrating force CF; R2 = 0.79 for content of sugar CS; R2 = 0.81 for pH) and relatively low average percent errors (ERR) (9.66%, 7.68% and 3.6% for CF, CS and pH, respectively). The quadratic polynomial step regression provides an accurate quality indices model, with high correlation (R2 = 0.92, 0.87, 0.83 for CF, CS and pH, respectively) between predicted and measured values and a relatively low error (5.47%, 3.45%, 2.57% for CF, CS and pH, respectively) for prediction. The feed-forward neural network also provides an accurate quality indices model with a high correlation (R2 = 0.90, 0.81, 0.87 for CF, CS and pH, respectively) between predicted and measured values and a relatively low average percent error (6.39%, 6.21%, 3.13% for CF, CS and pH, respectively) for prediction. These results prove that the electronic nose has the potential of becoming a reliable instrument to assess the peach quality indices.

Over the past years, electronic nose technology opened the possibility to exploit information on behavior aroma to assess fruit ripening stage. The objective in this study was to evaluate the capacity of electronic nose to monitoring the change in volatile production of mandarin during different picking-date, using a specific electronic nose device (PEN 2). Principal component analysis (PCA) and linear discriminant analysis (LDA) were used in order to investigate whether the electronic nose was able to distinguish among different picking-date (ripeness states). The loadings analysis was used to identify the sensors responsible for discrimination in the current pattern file. The results obtained prove that the electronic nose PEN 2 can discriminate successfully different picking-date on mandarin using LDA analysis. But, electronic nose was not able to detect a clear difference in volatile profile on mandarin using PCA analysis. During external validation using LDA was obtained to classified 92% of the total samples properly. Some sensors have the highest influence in the current pattern file for electronic nose PEN 2. A subset of few sensors can be chosen to explain all the variance. This result could be used in further studies to optimize the number of sensors.

An “electronic nose” has been used for the detection of adulterations of sesame oil. The system, comprising 10 metal oxide semiconductor sensors, was used to generate a pattern of the volatile compounds present in the samples. Prior to different supervised pattern recognition treatments, feature extraction techniques were employed to choose a set of optimal discriminant variables. Principal component analysis (PCA), Fisher linear transformation (FLT), stepwise linear discriminant analysis (Step-LDA), selection by Fisher weights (SFW) were used, respectively. And then, linear discriminant analysis (LDA), probabilistic neural networks (PNN), back propagation neural networks (BPNN) and general regression neural network (GRNN) were applied as pattern recognition techniques for the electronic nose. As for LDA and PNN, FLT was the most effective feature extraction method, while Step-LDA was the most effective way for BPNN and FLT was more suitable for GRNN. With only one sample misclassified in our experiment, LDA is more powerful than PNN. Excellent results were obtained in the prediction of percentage of adulteration in sesame oil by BPNN and GRNN. After training for some time, BPNN could predict the adulteration quantitatively more precisely than GRNN, whereas with FLT as its feature extraction method and without iterative training, GRNN could also yield rather acceptable results.

•Two-way MANOVA analysis studied the effect caused by storage conditions.•The flavor of mandarin was detected by E-nose and E-tongue simultaneously.•RF was applied for mandarin identification and quality changes prediction.•RF showed a perfect performance and can be an efficient method in this field.The aim of this work was to investigate the effects of storage temperature and time on internal quality of Satsuma mandarin (Citrus unshiu Marc.) by means of electronic nose (E-nose) and electronic tongue (E-tongue). The results obtained by two-way MANOVA analysis shows that 1) storage time and temperature did exist interaction effect on vitamin C and sugar/acid ratio, 2) no interaction was found on the content of total soluble solid, and 3) storage time showed no significant influence on the content of total acid. The fusion system, which was composed of signals from E-nose and E-tongue, based on random forest, predicted the internal quality parameters of mandarin fruits with fairly higher precision when compared with a single system. This work indicates that two-way MANOVA analysis might be employed to evaluate the loss of internal quality of fruits under different conditions, and the fusion system composed of E-nose and E-tongue can monitor mandarins at different storage conditions and predict internal quality changes successfully.Industrial relevanceStorage conditions for fresh fruit have been widely investigated, including shelf-life, storage temperature, waxing, and so on. Storage factors, during fresh fruit transportation or marketing, should be considered together and their interactions are also important. In this study, two-way MANOVA was applied to analyze the influence caused by storage time and temperature and also their interaction on the quality parameters. This paper describes the possible application of E-nose and E-tongue to trace the quality status of satsuma mandarin (Citrus unshiu Marc.) during different storage conditions. This technique could be used as a rapid and objective method applicable to routine quality control at any stage of the fruit supply chain.

An electronic nose (PEN 2) comprising ten metal oxide semiconductor sensors (W1C, W5S, W3C, W6S, W5C, W1S, W1W, W2S, W2W, W3S) were used to detection of five different stored duration wheats (wheats were harvested from 2000 to 2004, and named as W00, W01, W02, W03, and W04, respectively). A few of sensors were switched-off by multivariate analysis of variance and loading analysis. The responses signals of sensor W5C, W1S, W1W, W2S, W2W and W3S were chose for the pattern recognition. Principal component analysis (PCA) was applied to the signal of optimized sensor array, the five different stored duration of wheat were discriminated well and each group has strong convergence. The results obtained by network I (for optimized sensor array) presented the higher percent of correct classifications in comparison to network II (for original sensor array). The optimization of sensor array can improve the recognition performance of the electronic nose. The results obtained indicated that the electronic nose could discriminate successfully wheat of different age.

•Quality of raw fruit was traced by detecting the squeezed juices using an e-nose.•A novel semi-supervised classifier based on spectral clustering was applied.•The new classifier outperforms four supervised linear and nonlinear classifiers.•Regression models based on 20% and 70% of the whole dataset were compared.•Quality indices of cherry tomatoes were successfully predicted.An e-nose was presented to trace freshness of cherry tomatoes that were squeezed for juice consumption. Four supervised approaches (linear discriminant analysis, quadratic discriminant analysis, support vector machines and back propagation neural network) and one semi-supervised approach (Cluster-then-Label) were applied to classify the juices, and the semi-supervised classifier outperformed the supervised approaches. Meanwhile, quality indices of the tomatoes (storage time, pH, soluble solids content (SSC), Vitamin C (VC) and firmness) were predicted by partial least squares regression (PLSR). Two sizes of training sets (20% and 70% of the whole dataset, respectively) were considered, and R2 > 0.737 for all quality indices in both cases, suggesting it is possible to trace fruit quality through detecting the squeezed juices. However, PLSR models trained by the small dataset were not very good. Thus, our next plan is to explore semi-supervised regression methods for regression cases where only a few experimental data are available.

•Four batches of aged pecans are discriminated successfully using optimized array.•Non-searching feature selection method is used for sensor array optimization.•Three feature types are extracted to generate the initial feature matrix.•Sensors corresponding to the optimized feature matrix are chosen.In this research, an embedded metal oxide semiconductor (MOS) electronic nose (e-nose) was designed to detect Chinese pecan quality. To improve the performance of e-nose, three types of features were extracted to form initial feature matrix, including mean-differential coefficient value, stable value, and response area value. Furthermore, followed by the non-search feature selection strategy, optimized feature matrix was obtained through the procedure of mean analysis, variation coefficient analysis, cluster analysis and correlation analysis. It was observed that pecans were better classified after the optimization of initial feature matrix, shown by principal component analysis (PCA) score plot. And also the regression models of optimized feature matrix established by partial least squares regression (PLSR) (R2 = 0.9377) and back propagation neural networks (BPNN) (R2 = 0.9787) presented a better prediction capacity than these of initial one (PLSR: R2 = 0.8887; BPNN: R2 = 0.9093). In conclusion, the optimization method not only reduced data dimensionality but also improved electronic nose performance.

•An array of 18 metal-oxide based gas sensors was applied to trace peanuts quality.•The quality changes of peanuts during storage were detected by the sensor array.•The sensor signals were normalized in [0,1] and [−1,1] for quality tracing modeling.•PLSR and SVM were used to predict quality indices of peanuts and adulterated levels.•The prediction models on the basis of different normalized intervals were compared.Quality tracing models were set up for both unshelled peanuts and peanut kernels by applying an array of 18 metal-oxide (MOX) based gas sensors. Acid value, peroxide value and content of crude fat of the peanuts at different storage times were measured by traditional methods as a reference. Classification results for both unshelled peanuts and peanut kernels at different storage times based on Linear Discriminant Analysis (LDA) and Support Vector Machine (SVM) were acceptable Storage time, acid value, peroxide value and content of crude fat of peanuts were predicted by Partial Least Squares Regression (PLSR) and SVM on the basis of different normalized datasets. Original datasets, datasets normalized in [0,1] and in [−1,1] were considered. PLSR and SVM provided better prediction results when normalized in [0,1] and [−1,1], respectively. Correlations between adulterated levels (stale peanuts blended in fresh peanuts at levels of 0%, 25%, 50%, 75% and 100%) and sensor signals were researched by PLSR and SVM. It was found that the sensor signals and adulterated levels exhibited good correlation (R2 > 0.801 for training and testing sets by both methods). Meanwhile, The R2 for training and testing sets were 0.941 and 0.896 by applying SVM, respectively, and both of them were correspondingly higher than the R2 for training and testing sets by PLSR (training: R2 = 0.812; testing: R2 = 0.802). The research indicates that the 18 MOX based gas sensors combined with appropriate chemometrics methods can be used as a non-destructive method in detecting peanut quality.

•The manufacture and storage processes of set yogurt were monitored by VE-tongue.•Two types of chronoamperometric current methods, MRPV and MSPV, were employed.•The visualization of PCM analyzed the correlation of the combination of EDs and PWs.•The changing of samples in each process was exhibited clearly by chemometrics methods.A voltammetric electronic tongue (VE-tongue) was self-developed to monitor the fermentation, post-ripeness and storage processes of set yogurt. Multifrequency rectangular pulse voltammetry (MRPV) and multifrequency staircase pulse voltammetry (MSPV) were applied as potential waveforms, and the ‘area method’ was applied to extract feature data from the original responses. The ANOVA analysis was performed to evaluate the effects of electrodes and potential waveforms on the electrochemical responses, and the visualization of the Pearson Correlation Coefficient was performed to analyze the correlations among each combination of electrode and potential waveform. Discriminant function analysis worked better than principal component analysis based on the fusion data of MRPV- and MSPV-area data in classifying set yogurts in the stages of fermentation, post-ripeness and storage, respectively. VE-tongue with support vector machine (SVM) worked efficiently and stably in predicting the fermentation time, pH and viscosity of samples during the fermentation process; while the use of VE-tongue and partial least squares regression (PLSR) should be the first choice for the prediction work during the post-ripeness and storage processes. According to the classification and prediction results, the VE-tongue with different potential waveforms was proved a promising tool for monitoring the fermentation, post-ripeness and storage processes of set yogurt.

In this paper, responses of sensor array in electronic nose were employed to establish quality indices model able to describe the different picking dates of “xueqing” pear. The multivariate calibration methods, multiple linear regression (MLR), principal component regression (PCR) and partial least-squares regressions (PLS) were applied to predict the quality indices of “xueqing” pear from different picking dates based on the signal of electronic nose. All models for firmness and soluble solids content show a good prediction performance. However the acidity, there was a very poor correlation with the signal of the electronic nose. It was found that MLR led to more precise predictions than the other multivariate calibration methods. The results indicate that it is possible to use this non-destructive technique for measuring “xueqing” pear quality characteristics. The methods have the potential to estimate chemical and physical properties of pear from signal of electronic nose.

BackgroundAmerican ginseng (Panax quinquefolius L.) and Asian ginseng (Panax ginseng Meyer) products, such as slices, have a similar appearance, but they have significantly different prices, leading to widespread adulteration in the commercial market. Their aroma characteristics are attracting increasing attention and are supposed to be effective and nondestructive markers to determine adulteration.MethodsThe aroma characteristics of American and Asian ginseng were investigated using gas chromatography–mass spectrometry(GC-MS) and an electronic nose (E-nose). Their volatile organic compounds were separated, classified, compared, and analyzed with different pattern recognition.ResultsThe E-nose showed a good performance in grouping with a principle component analysis explaining 94.45% of variance. A total of 69 aroma components were identified by GC-MS, with 35.6% common components and 64.6% special ingredients between the two ginsengs. It was observed that the components and the number of terpenes and alcohols were markedly different, indicating possible reasons for their difference. The results of pattern recognition confirmed that the E-nose processing result is similar to that of GC-MS. The interrelation between aroma constituents and sensors indicated that special sensors were highly related to some terpenes and alcohols. Accordingly, the contents of selected constituents were accurately predicted by corresponding sensors with most R2 reaching 90%.ConclusionCombined with advanced chemometrics, the E-nose is capable of discriminating between American and Asian ginseng in both qualitative and quantitative angles, presenting an accurate, rapid, and nondestructive reference approach.

•An E-nose was used to characterize five types of strawberry juices based on processing approaches.•ELM was first used in the field of E-nose data processing.•Quality indices tested by conventional methods were predicted by MLR, PLS, Lib-SVM and ELM.•ELM performed best both in the classification and regression.An electronic nose (E-nose) has been used to characterize five types of strawberry juices based on different processing approaches (i.e., Microwave Pasteurization, Steam Blanching, High Temperature Short Time Pasteurization, Frozen–Thawed, and Freshly Squeezed). Juice quality parameters (vitamin C and total acid) were detected by traditional measuring methods. Multivariate statistical methods (Principle Component Analysis, Linear Discriminant Analysis, Multiple Linear Regression, and Partial Least Squares Regression) and neural networks (Extreme Learning Machine (ELM), Learning Vector Quantization and Library Support Vector Machines) were employed for qualitative classification and quantitative regression. ELM showed best performances on classification and regression, indicating that ELM would be a good choice for E-nose data treatment. Results provide promising principles for the elaboration of E-nose which could be used to discriminate processed juices and to predict juice quality parameters based on appropriate algorithms for the beverage industry.

•The quality of Citrus was detected by E-nose and E-tongue simultaneously.•ELM as a novel data mining method was first used in fusion of E-nose and E-tongue.•SVM, RF, and ELM were compared with accuracy rate and regression parameters.•E-nose and E-tongue with RF or ELM could be an alternative to monitor fruit quality.This paper demonstrates a joint way employing both of an electronic nose (E-nose) and an electronic tongue (E-tongue) to discriminate two types of satsuma mandarins from different development stages and to trace the internal quality changes (i.e. ascorbic acid, soluble solids content, total acid, and sugar/acid ratio). Extreme Learning Machine (ELM), Random Forest (RF) and Support Vector Machine (SVM) were applied for qualitative classification and quantitative prediction. The models were compared according to accuracy rate and regression parameters. For classification, the three systems (E-nose, E-tongue, and the fusion system) achieved perfect results respectively. For internal quality prediction, the RF and ELM models obtained better performance than the SVM models. The fusion systems had an advantage when compared with the signal system. This study shows that the E-nose and E-tongue systems combined with RF or ELM could be a fast and objective detection system to trace fruit internal quality changes.

Fast recognition and characterization of preserved licorice apricot were studied by electronic tongue. The E-tongue signals were analyzed by pattern recognition techniques. Five brands of preserved licorice apricot were discriminated with strong convergence by pattern recognition techniques, Principal Component Analysis, Canonical discriminant Analysis and Cluster Analysis. The characterization of the samples obtained by Back-Propagation Neural Network (BPNN) and Partial Least Squares regression (PLSR) were 100% accurate both for training and test set, and the highest correlation between observed and predicted values was obtained for aerobic plate count, (0.9943, 0.9951) followed by total sugar content (0.9941, 0.9853), content of sodium chloride (0.9926, 0.9902), sulfur dioxide residues (0.9894, 0.9928) with BPNN method. All pattern recognition methods performed for the characterization and classification showed the potential of E-tongue as a rapid tool in the analysis and characterization of preserved fruits.► The taste of preserved licorice apricot was studied by an electronic tongue. ► Pattern recognition methods were employed to classify the groups of samples. ► Physicochemical parameters were well predicted by E-tongue responses. ► A rapid quality analysis method was reported for preserved fruit based on E-tongue.

The profiles of volatile compounds emitted by plants varies in response to damage or herbivore attack. The potential of electronic nose technology to monitor such changes, with the aim of diagnosing plant health was investigated. An electronic nose (E-nose) was used to analyse rice plants that were subjected to different types of treatments causing damage, and the results were compared to those of undamaged control plants. Principal component analysis (PCA), linear discrimination analysis (LDA), cluster analysis (CA), back-propagation neural network (BPNN), and learning vector quantisation (LVQ) network were used to evaluate the E-nose data. The results indicated that the E-nose can successfully discriminate between rice plants with different types of damage. The discrimination was more pronounced after the LDA than after the PCA. The front 5 principal component values of the PCA were extracted and they acted as the input date for the neural network analyses. Good discrimination results were obtained using these front 5 principal component values in LVQ and BPNN. The results demonstrated that it is plausible to use E-nose technology as a method for monitoring rice cultivation practices.Highlights►Potential of electronic nose technology to monitor such changes for plant health diagnosis. ►E-nose was used to discriminate rice plants subjected to different types of damage compared with undamaged control plants. ►The results indicate that the E-nose can successfully discriminate rice plants with different types of damage. ►Good discrimination results are obtained using the front five principal component values by linear discrimination analysis and back-propagation neural network.

Few researches have been reported on the long-term germination characteristics and the effect of high gamma radiation dose on cereal seeds. In this paper, to observe the effects of gamma irradiation (0–3 kGy) on the germination of wheat seed in long-term (within 20 months), wheat seed was dried after irradiation and the germination experiment during storage time was conducted. It was found that the lengths of buds of irradiated wheat seeds diminished, the roots of irradiated wheat seeds disappeared, and no germinations in irradiated wheat seed was found. The influence of γ-ray irradiation on roots was more significant than that on buds. After long-term storage, the germination of irradiated wheat seeds increased.Highlights► We observe the effects of irradiation on the germination of wheat seed in long-term. ► Germination of irradiated wheat seeds increased after long-term storage. ► Influence of irradiation on roots was more significant than buds.

Electronic nose technology offers non-destructive alternative to sense aroma, can be used to assess fruit ripening stage during shelf life. The objective of this study was to monitor tomato storage shelf life during two storage treatments using PEN 2 electronic nose (E-nose). Principal component analysis (PCA) and linear discriminant analysis (LDA) were used to distinguish the different tomato storage time. The obtained results proved that tomato with different storage time can be monitored by the E-nose, but very clear separation among all groups of different storage time was not achieved. By PCA and LDA, E-nose could more clearly discrimination storage time of tomato in carton box than one in folded bag. The correlations between the measured and predicted values of fruit quality attribute (soluble solids content, pH, and puncture force) showed poor prediction performance on the base of signals of E-nose sensors.

Wheats of five storage ages and with 15 degrees of insect damage were evaluated and classified by the static-headspace sampling method using an electronic nose (E-nose). A commercial E-nose (PEN2) comprising 10 metal-oxide semiconductor (MOS) sensors was used to generate a typical chemical fingerprint of the volatile compounds present in the samples. Principal-component analysis (PCA) and linear-discriminant analysis (LDA) were applied to the generated patterns to achieve classification into the five groups of different storage-age wheats and the 15 groups of different degrees of insect-damaged wheat. The results obtained indicated that the E-nose could discriminate successfully among wheats of different age and with different degrees of insect damage.

The freshness of fruit is relatively easy to authenticate by its morphological characteristics, while processing fruits into juices makes the freshness difficult to identify. In this paper, cherry tomatoes at different storage temperatures (4 and 25 °C) and shelf lives (SLs, 16 days at 4 °C and 8 days at 25 °C) were squeezed for use in 100% juices. Quality indices (SL, pH, soluble solids content (SSC), vitamin C (VC) concentration and firmness) of these cherry tomatoes were determined through analysing the juices using two sensor systems – an e-nose and an e-tongue. Support vector regression (SVR) was applied to predict the quality indices. The prediction performances based on a one sensor system, as well as a combination of two systems, were compared. The results showed that the e-tongue, which presents a similar prediction performance to the combination system, presents a better prediction performance (with higher squared correlation coefficients (R2) and a lower standard error of prediction (SEP)) than the e-nose. For tomatoes stored at 4 °C, the prediction parameters (R2, SEP) based on the e-tongue data for the SL, pH, SSC, VC concentration and firmness are (0.998, 0.295 d), (0.971, 0.022), (0.906, 0.075 °Brix), (0.978, 1.005 mg per 100 g) and (0.906, 0.292 N), respectively. For tomatoes stored at 25 °C, the prediction parameters (R2, SEP) based on the e-tongue data for the SL, pH, SSC, VC concentration and firmness are (0.997, 0.193 d), (0.934, 0.017), (0.957, 0.075 °Brix), (0.902, 0.897 mg per 100 g) and (0.908, 0.593 N), respectively. These results prove that it is possible to measure the freshness of fruits that are squeezed for juice consumption using sensor systems, and that the combination of sensor systems is not always better than using a one sensor system.