Discrimination among Fresh, Frozen-Stored and Frozen-Thawed Beef Cuts by Hyperspectral Imaging

The detection of the storage state of frozen meat, especially meat frozen-thawed several times, has always been important for food safety inspections. Hyperspectral imaging (HSI) is widely applied to detect the freshness and quality of meat or meat products. This study investigated the feasibility of the low-cost HSI system, combined with the chemometrics method, to classify beef cuts among fresh (F), frozen-stored (F-S), frozen-thawed three times (F-T-3) and frozen-thawed five times (F-T-5). A compact, low-cost HSI system was designed and calibrated for beef sample measurement. The classification model was developed for meat analysis with a method to distinguish fat and muscle, a CARS algorithm to extract the optimal wavelength subset and three classifiers to identify each beef cut among different freezing processes. The results demonstrated that classification models based on feature variables extracted from differentiated tissue spectra achieved better performances, with ACCs of 92.75% for PLS-DA, 97.83% for SVM and 95.03% for BP-ANN. A visualization map was proposed to provide detailed information about the changes in freshness of beef cuts after freeze-thawing. Furthermore, this study demonstrated the potential of implementing a reasonably priced HSI system in the food industry.

Inversion of soil water and salt information based on UAV hyperspectral remote sensing and machine lear-ning

Accurate diagnosis of water and salt information in saline agricultural lands is crucial for long-term soil quality improvement and arable land conservation. In this study, we extracted field-scale vegetation canopy spectral information by UAV hyperspectral information, transforming the reflectance (R) to standard normal variate transformation (SNV), multiplicative scatter correction (MSC), first derivative of reflectance (FDR) and second derivative of reflectance (SDR). We determined the optimal spectral transformation forms of soil water content (SWC), soil pH, and soil salt content (SSC) by the maximum absolute correlation coefficient (MACC), and extracted the feature bands by competitive adaptive reweighted sampling (CARS). We constructed an inversion model of soil water and salt information by partial least squares regression (PLSR), random forest (RF), and extreme gradient boosting (XGBoost). The results showed that R, FDR and MSC were the best spectral transformation types for soil water content, soil pH, and soil salt content, and the corresponding MACC were 0.730, 0.472 and 0.654, respectively. The CARS algorithm effectively eliminated the irrelevant variables, optimally selecting 16-17 feature bands from 150 spectral bands. Both soil water content and soil pH performed best with XGBoost model, achieving determination coefficient of validation (Rp2) 0.927 and 0.743, and the relative percentage difference (RPD) amounted to 3.93 and 2.45. For soil salt content, the RF model emerged as the best inversion method with Rp2 and RPD of 0.427 and 1.64, respectively. The study could provide a reference solution for the integrated remote sensing monitoring of soil water and salt information in space and sky, serving as a scientific guide for the amelioration and sustainable management of saline lands.

Rapid Determination of Cellulose and Hemicellulose Contents in Corn Stover Using Near-Infrared Spectroscopy Combined with Wavelength Selection

The contents of cellulose and hemicellulose (C and H) in corn stover (CS) have an important influence on its biochemical transformation and utilization. To rapidly detect the C and H contents in CS by near-infrared spectroscopy (NIRS), the characteristic wavelength selection algorithms of backward partial least squares (BIPLS), competitive adaptive reweighted sampling (CARS), BIPLS combined with CARS, BIPLS combined with a genetic simulated annealing algorithm (GSA), and CARS combined with a GSA were used to select the wavelength variables (WVs) for C and H, and the corresponding regression correction models were established. The results showed that five wavelength selection algorithms could effectively eliminate irrelevant redundant WVs, and their modeling performance was significantly superior to that of the full spectrum. Through comparison and analysis, it was found that CARS combined with GSA had the best comprehensive performance; the predictive root mean squared errors of the C and H regression model were 0.786% and 0.893%, and the residual predictive deviations were 3.815 and 12.435, respectively. The wavelength selection algorithm could effectively improve the accuracy of the quantitative analysis of C and H contents in CS by NIRS, providing theoretical support for the research and development of related online detection equipment.

[Rapid determination of active components in Ginkgo biloba leaves by near infrared spectroscopy combined with genetic algorithm joint extreme learning machine]

Near-infrared spectroscopy(NIRS) combined with band screening method and modeling algorithm can be used to achieve the rapid and non-destructive detection of the traditional Chinese medicine(TCM) production process. This paper focused on the ginkgo leaf macroporous resin purification process, which is the key technology of Yinshen Tongluo Capsules, in order to achieve the rapid determination of quercetin, kaempferol and isorhamnetin in effluent. The abnormal spectrum was eliminated by Mahalanobis distance algorithm, and the data set was divided by the sample set partitioning method based on joint X-Y distances(SPXY). The key information bands were selected by synergy interval partial least squares(siPLS); based on that, competitive adaptive reweighted sampling(CARS), successive projections algorithm(SPA) and Monte Carlo uninformative variable(MC-UVE) were used to select wavelengths to obtain less but more critical variable data. With selected key variables as input, the quantitative analysis model was established by genetic algorithm joint extreme learning machine(GA-ELM) algorithm. The performance of the model was compared with that of partial least squares regression(PLSR). The results showed that the combination with siPLS-CARS-GA-ELM could achieve the optimal model performance with the minimum number of variables. The calibration set correlation coefficient R_c and the validation set correlation coefficient R_p of quercetin, kaempferol and isorhamnetin were all above 0.98. The root mean square error of calibration(RMSEC), the root mean square error of prediction(RMSEP) and the relative standard errors of prediction(RSEP) were 0.030 0, 0.029 2 and 8.88%, 0.041 4, 0.034 8 and 8.46%, 0.029 3, 0.027 1 and 10.10%, respectively. Compared with the PLSR me-thod, the performance of the GA-ELM model was greatly improved, which proved that NIRS combined with GA-ELM method has a great potential for rapid determination of effective components of TCM.

Rapid authentication of Pseudostellaria heterophylla (Taizishen) from different regions by near-infrared spectroscopy combined with chemometric methods

Pseudostellaria heterophylla is a very popular traditional Chinese medicine herb, also called "Taizishen." Discrimination of P. heterophylla from different regions is critical for ensuring the effectiveness of drug use, because the drug effects of P. heterophylla from different regions are diversity of each other. To discriminate P. heterophylla from different regions rapidly and effectively, a model extracted by competitive adaptive reweighted sampling (CARS) was established. Original spectra of P. heterophylla in wave number range of 10,000 to 4,000 cm^-1 were acquired. Orthogonal partial least squares discriminant analysis (OPLS-DA) was also used to establish a suitable model. CARS was performed for extracting key wave number variables. We found that the near-infrared spectrum of a series of samples analyzed by Row-center-SG, CARS, and OPLS-DA can effectively distinguish the P. heterophylla from different regions, and the accuracy of OPLS-DA model is also satisfactory in terms of good discrimination rate. These results show that the Row-center-SG, CARS, and OPLS-DA model can be used to identify the P. heterophylla from different regions. PRACTICAL APPLICATION: According to our research results, we can draw a conclusion that our research results may be used to distinguish the traditional Chinese medicine from those from different places of origin and the powder with similar appearance.

Determination of metmyoglobin in cooked tan mutton using Vis/NIR hyperspectral imaging system

In this study, the ENVI 4.6 software was used to obtain the spectral reflection value of samples. The outlier samples were eliminated by the Monte Carlo method, and then SPXY (sample set partitioning based on be x-y distances) was used to divide the calibration set and prediction set. The spectral images were pretreated and characteristic wavelengths were extracted. The spectral models of full and pretreated spectra and characteristic bands were established by partial least squares regression (PLSR) and principle component regression (PCR), and the optimal modeling combination was selected. The results showed that the modeling effect of the original spectrum was the best. In full-PLSR model, the determination coefficient of the calibration set (Rc² ), the determination coefficient of prediction set (Rp² ), and the determination coefficient of interactive verification set (Rcv² ) were 0.8804, 0.7375, and 0.7422, and root-mean-square error of calibration set (RMSEC), root-mean-square error of prediction (RMSEP), and root mean square error of interactive validation set (RMSECV) were 2.3630, 2.9607, and 3.4209, respectively. PLSR and PCR models were established to obtain the optimal models of CARS-PLSR and PCR-PLSR. In the CARS-PLSR model, the Rc² , Rp² , and Rcv² were 0.9135, 0.7654, and 0.8171, respectively, while RMSEC, RMSEP, and RMSECV were 2.0275, 2.9306, and 2.9262, respectively. In the iRF-PCR model, Rc² , Rp² , and Rcv² were 0.7952, 0.7372, and 0.7280, respectively, while RMSEC, RMSEP, and RMSECV were 3.0207, 2.8278, and 3.4288, respectively. This study has demonstrated that visible and near-infrared hyperspectral imaging system can rapidly predict the content of metmyoglobin in cooked tan mutton. PRACTICAL APPLICATION: This study has demonstrated that visible and near-infrared (Vis/NIR) hyperspectral imaging system can rapidly predict the content of MetMb in cooked tan mutton. With the advantages of nondestructive, rapid, real-time, Vis/NIR, hyperspectral imaging system can be widely expanded and applied to the detection of myoglobin in meat to evaluate the color of meat.

Optimizing Rice Near-Infrared Models Using Fractional Order Savitzky-Golay Derivation (FOSGD) Combined with Competitive Adaptive Reweighted Sampling (CARS)

Developing a rapid and stable method for analyzing the quality parameters of rice is important. Near-infrared (NIR) spectroscopy combined with chemometric techniques have been used to predict the critical contents of rice and shown its accuracy and stability. To further improve the predictive ability, we combine the derivative method of fractional order Savitzky-Golay derivation (FOSGD) with the wavelength selection method of competitive adaptive reweighted sampling (CARS). Compared with the traditional integer order Savitzky-Golay derivation (IOSGD), the FOSGD could improve the resolution ratio of the raw spectra more effectively. The wavelength selection method, CARS, could further extract the informative variables from the processed spectra. Four key contents of rice samples, including moisture, amylose, chalkiness degree, and gel consistency, were utilized to validate this method. The prediction results indicated that partial least squares (PLS) models optimized with FOSGD-CARS own higher accuracy and stability with smaller the root mean squared error of cross validations (RMSECVs) and root mean squared error of predictions (RMSEPs). The proposed method is convenient and provides a practical alternative for rice analysis.

A Rapid and Nondestructive Method for Simultaneous Determination of Aflatoxigenic Fungus and Aflatoxin Contamination on Corn Kernels

Conventional methods for detecting aflatoxigenic fungus and aflatoxin contamination are generally time-consuming, sample-destructive, and require skilled personnel to perform, making them impossible for large-scale nondestructive screening detection, real-time, and on-site analysis. Therefore, the potential of visible-near-infrared (Vis-NIR) spectroscopy over the 400-2500 nm spectral range was examined for determination of aflatoxigenic fungus infection and the corresponding aflatoxin contamination on corn kernels in a rapid and nondestructive manner. The two A. flavus strains, AF13 and AF38, were used to represent the aflatoxigenic fungus and nonaflatoxigenic fungus, respectively, for artificial inoculation on corn kernels. The partial least-squares discriminant analysis (PLS-DA) models based on different combinations of spectral range (I: 410-1070 nm; II: 1120-2470 nm), corn side (endosperm or germ side), spectral variable number (full spectra or selected variables), modeling approach (two-step or one-step), and classification threshold (20 or 100 ppb) were developed and their performances were compared. The first study focusing on detection of aflatoxigenic fungus-infected corn kernels showed that, in classifying the "control+AF38-inoculated" and AF13-inoculated corn kernels, the full spectral PLS-DA models using the preprocessed spectra over range II and one-step approach yielded more accurate prediction results than using the spectra over range I and the two-step approach. The advantage of the full spectral PLS-DA models established using one corn side than the other side were not consistent in the explored combination cases. The best full spectral PLS-DA model obtained was obtained using the germ-side spectra over range II with the one-step approach, which achieved an overall accuracy of 91.11%. The established CARS-PLSDA models performed better than the corresponding full-spectral PLS-DA models, with the better model achieved an overall accuracy of 97.78% in separating the AF13-inoculated corn kernels and the uninfected control and AF38-inoculated corn kernels. The second study focusing on the detection of aflatoxin-contaminated corn kernels showed that, based on the aflatoxin threshold of 20 and 100 ppb, the best overall accuracy in classifying the aflatoxin-contaminated and healthy corn kernels attained 86.67% and 84.44%, respectively, using the CARS-PLSDA models. The quantitative modeling results using partial least-squares regression (PLSR) obtained the correlation coefficient of prediction set ( R_P) of 0.91, which indicated the possibility of using Vis-NIR spectroscopy to quantify aflatoxin concentration in aflatoxigenic fungus-infected corn kernels.