Ernest Teye

Rapid differentiation of Ghana cocoa beans by FT-NIR spectroscopy coupled with multivariate classification.

Quick, accurate and reliable technique for discrimination of cocoa beans according to geographical origin is essential for quality control and traceability management. This current study presents the application of Near Infrared Spectroscopy technique and multivariate classification for the differentiation of Ghana cocoa beans. A total of 194 cocoa bean samples from seven cocoa growing regions were used. Principal component analysis (PCA) was used to extract relevant information from the spectral data and this gave visible cluster trends. The performance of four multivariate classification methods: Linear discriminant analysis (LDA), K-nearest neighbors (KNN), Back propagation artificial neural network (BPANN) and Support vector machine (SVM) were compared. The performances of the models were optimized by cross validation. The results revealed that; SVM model was superior to all the mathematical methods with a discrimination rate of 100% in both the training and prediction set after preprocessing with Mean centering (MC). BPANN had a discrimination rate of 99.23% for the training set and 96.88% for prediction set. While LDA model had 96.15% and 90.63% for the training and prediction sets respectively. KNN model had 75.01% for the training set and 72.31% for prediction set. The non-linear classification methods used were superior to the linear ones. Generally, the results revealed that NIR Spectroscopy coupled with SVM model could be used successfully to discriminate cocoa beans according to their geographical origins for effective quality assurance.

Nondestructive detection of fish freshness during its preservation by combining electronic nose and electronic tongue techniques in conjunction with chemometric analysis

A new method was developed to detect fish freshness nondestructively by combining electronic nose (E-nose) and electronic tongue (E-tongue) in conjunction with chemometric methods. An E-nose with nine metal oxide semiconductor gas sensors and a commercial E-tongue were employed in this research. Pseudosciaena crocea stored at 4 °C for different days were used as experimental samples. Total viable counts (TVC) of the fish were detected by the conventional method. E-nose and E-tongue data were analyzed by principal component analysis. Three-layer radial basis function neural network (RBF-NN) models were established for qualitative discrimination of the fish freshness. Performances of RBF-NN models with different numbers of principal components (PCs) as the input were compared. Experimental results revealed that the best RBF-NN model was acquired at seven PCs of E-nose data with an optimal performance of 87.9% and 80.0% in the training set and prediction set respectively. While, the best RBF-NN model of E-tongue data analysis was at five PCs with an optimal performance of 86.3% in the training set and 81.8% in prediction set. Another RBF-NN model was built with the combination of E-nose and E-tongue. The result shows that the discrimination rates improved to 94.0% and 93.9% in the training set and prediction set respectively. A support vector machine regression model was applied to establish a relationship between the combined data from E-nose and E-tongue and from TVC values for quantitative determination. A high correlation was found between the merged data and the parameter of TVC with correlation coefficients more than 0.91. The results proved that, a single system of E-nose and E-tongue was enough to classify samples stored on different days at 4 °C, while a higher discrimination rate was acquired by the combination of the two sensors. The combined system could also be used to quantitatively evaluate the fish freshness. In conclusion, the combined system of E-nose and E-tongue in conjunction with appropriate chemometric analysis methods can conveniently and nondestructively evaluate the freshness of fish stored at 4 °C.

Estimating cocoa bean parameters by FT-NIRS and chemometrics analysis

Rapid analysis of cocoa beans is an important activity for quality assurance and control investigations. In this study, Fourier transform near infrared spectroscopy (FT-NIRS) and chemometric techniques were attempted to estimate cocoa bean quality categories, pH and fermentation index (FI). The performances of the models were optimised by cross-validation and examined by identification rate (%), correlation coefficient (Rpre) and root mean square error of prediction (RMSEP) in the prediction set. The optimal identification model by back propagation artificial neural network (BPANN) was 99.73% at 5 principal components. The efficient variable selection model derived by synergy interval back propagation artificial neural network regression (Si-BPANNR) was superior for pH and FI estimation. Si-BPANNR model for pH was Rpre=0.98 and RMSEP=0.06, while for FI was Rpre=0.98 and RMSEP=0.05. The results demonstrated that FT-NIRS together with BPANN and Si-BPANNR model could successfully be used for cocoa beans examination.

Novel Prediction of Total Fat Content in Cocoa Beans by FT-NIR Spectroscopy Based on Effective Spectral Selection Multivariate Regression

Total fat content is a major quality parameter that chocolate manufactures consider when selecting cocoa beans. This paper attempted the feasibility of measuring total fat content in cocoa beans by using Fourier transform near-infrared (FT-NIR) spectroscopy based on a novel systematic study on efficient spectral variables selection multivariate regression. After the efficient spectra interval selection by synergy interval partial least squares (Si-PLS), the data were treated with support vector machine regression (SVMR) leading to synergy interval support vector machine regression (Si-SVMR). Experimental results showed that the model based on the novel Si-SVMR algorithm was superior to the others. The optimum results were assessed by root-mean-square error of prediction (RMSEP) and correlation coefficient (Rpre) in the prediction set. The performance of Si-SVMR model was RMSEP = 0.015 and Rpre = 0.9708. This study has demonstrated that the total fat content in cocoa beans could rapidly be predicted by FT-NIR spectroscopy and Si-SVMR technique. The novel strength and accuracy of Si-SVMR in contrast to other multivariate algorithms has been derived.

Non-destructive evaluation of total volatile basic nitrogen (TVB-N) and K-values in fish using colorimetric sensor array

For rapid evaluation of fish freshness, a colorimetric sensor array has been developed for the sensitive detection to measure simultaneously TVB-N and K value of fish during its storage period. Silver carps were taken as fish samples which were stored at constant temperature of 4 °C during experiment period. 10 kinds of porphyrin compounds and 6 pH indicators were selected as chromogenic materials in this experiment according to the previous study and the theoretical research. For comparison, total volatile basic nitrogen (TVB-N) values of fishes were tested by conventional chemical method, and the K-values were measured using High Performance Liquid Chromatography (HPLC). As sensing materials used in the sensor array were chromogenic, the color of the sensor array changed when reacting with odor emitted by fish sample. The color change profiles of the sensor array before and after exposure to the odor of each sample were got using image processing method. And color features were extracted to be analyzed using principal component analysis (PCA), linear discriminant analysis (LDA). The relationship between these analysis results and the TVB-N values and K-values obtained by conventional methods were established using support vector regression (SVR). And therefore models were set up for rapid prediction of TVB-N values and K-values, respectively. For the SVR model of TVB-N content and K-values, calibration correlation coefficient (Rtr) was 0.8564 and 0.8712, and the root mean square error of calibration (RMSEC) was 4.2177 and 0.06127, respectively. It is feasible to predict TVB-N values and K-values according to experiment results of colorimetric sensor array. The results indicated that the novel method based on colorimetric sensor array developed provide a feasible way for rapid and nondestructive evaluation of fish freshness.

Rapid measurement of total polyphenols content in cocoa beans by data fusion of NIR spectroscopy and electronic tongue

Total polyphenols content (TPC) is an important measure of phytochemicals in cocoa beans due to its numerous health benefits. This work attempts to measure the total polyphenols content in cocoa beans by using a novel approach of integrating near infrared spectroscopy (NIRS) and electronic tongue (ET). 110 samples of cocoa beans with different polyphenol content were used for data acquisition by NIRS and ET. The optimum individual characteristic variables were extracted and scaled by normalization in principal component analysis (PCA). Support vector machine regression (SVMR) was used to construct the model. The performance of the final model was evaluated according to: correlation coefficient (Rpre), root mean square error of prediction (RMSEP) and bias in the prediction set. Compared with a single technique (NIRS or ET), the data fusion was superior for the determination of TPC in cocoa beans. The optimal data fusion model was achieved with: Rpre = 0.982, RMSEP = 0.900 g g−1 and bias = 0.013 in the prediction set. The overall results demonstrate that integrating NIRS and ET is possible and could improve the prediction of TPC in cocoa beans.

Simultaneous Measurement of Titratable Acidity and Fermentation Index in Cocoa Beans by Electronic Tongue Together with Linear and Non-linear Multivariate Technique

Titratable acidity (TA) and fermentation index (FI) are important quality indicators of cocoa beans. This paper attempted the simultaneous analysis of these indicators by electronic tongue (ET) and two multivariate calibrations. ET was used for data acquisition, while partial least squares (PLSs) and principal component support vector machine regression (PC-SVMR) were used to build the calibration models. Some parameters were optimized simultaneously by leave-one-out cross-validation (LOOCV) in calibrating the model. The performance of the model was tested according to root mean square error of prediction (RMSEP) and correlation coefficient (Rpre) in the prediction set. The results revealed that PC-SVMR model was superior to PLS model in this work. The optimal PC-SVMR model for TA was Rpre = 0.960 and RMSEP = 0.0077, while for FI, this was Rpre = 0.954 and RMSEP = 0.058. This study demonstrated that ET together with SVMR could be used to analyze titratable acidity and fermentation index in cocoa beans for quality control purposes.

Nondestructive determination of bamboo shoots lignification using FT-NIR with efficient variables selection algorithms

To achieve the rapid and nondestructive determination of bamboo shoots lignification associated with crude fiber content and firmness, Fourier transform near infrared (FT-NIR) spectroscopy technique was used in this paper. The identification of efficient spectral variables selection algorithms: backward interval PLS (Bi-PLS), Monte Carlo uninformative variables elimination method (MC-UVE), competitive adaptive reweighted sampling (CARS) and genetic algorithms (GA) were also discussed. The partial least squares (PLS) algorithm was applied to establish prediction models for crude fiber content and firmness after spectral preprocessing and variables selection. The correlation coefficient (R) and root mean square error of prediction (RMSEP) were used to assess predictive effects of PLS models. Modeling results showed that the CARS-GA-PLS model was prime for crude fiber content prediction (R = 0.9508, RMSEP = 0.0598), and the CARS-PLS model was superior for firmness prediction (R = 0.9681, RMSEP = 0.8003). The overall results sufficiently demonstrated that the FT-NIR spectroscopy technique could determine successfully crude fiber content and firmness of postharvest bamboo shoots.