Research Progress of Near-Infrared Fluorescence Immunoassay.

Near-infrared fluorescence probes (NIFPs) have been widely used in immunoassay, bio-imaging and medical diagnosis. We review the basic principles of near-infrared fluorescence and near-infrared detection technology, and summarize structures, properties and characteristics of NIFPs (i.e., cyanines, xanthenes fluorescent dyes, phthalocyanines, porphyrin derivates, single-walled carbon nanotubes (SWCNTs), quantum dots and rare earth compounds). We next analyze applications of NIFPs in immunoassays, and prospect the application potential of lateral flow assay (LFA) in rapid detection of pathogens. At present, our team intends to establish a new platform that has highly sensitive NIFPs combined with portable and simple immunochromatographic test strips (ICTSs) for rapid detection of food-borne viruses. This will provide technical support for rapid detection on the port.

[Identification of Fertilized Chicken Eggs Based on Visible/Near-Infrared Spectrum During Early Stage of Incubation].

In the process of chicken egg hatching, some eggs can not be hatched successfully due to the absence of fertilization. These eggs not only cause a lot of waste, but also infect other normal eggs with bacteria. In the study, the fertilized eggs and clear eggs is identified by using the visible/near-infrared spectrum. It is of great necessity to get the best time of identifying the clear eggs in the early of hatching, so the variation of eggs' quality in the condition of hatching over time is studied. The results show that eggs are fresh after 24 hours' hatching and eggs can not be eaten after 72 hours' hatching while the best time of identification is within 36 hours. Static acquisition system is developed based on visible/near-infrared transmission spectrum for acquiring spectrum. Comparing the effect of the model of the different samples of same breed and samples of different breed, the different part of spectrum among fertilized eggs and clear eggs is deleted which caused by the color of eggshell and yolk, the effective spectral band are 355～590 and 670～1 025 nm. Adopting the pretreatment of PCA and comparing the accuracy of the various mathematical models with different time and the number of principal components decide the best number of principal components. Considering the production efficiency and comparing the different pretreatment methods of spectrum, for examples, SNV, MSC, Derivative correction and PCA, and various mathematical models are combined to establish the most efficient discriminant model. The result shows that the most efficient discriminant model is established with Fisher and based on the pretreatment of PCA after 24 hours' hatching. And the precision rate is 87.18%. The study provides a new way for nondestructive and online identification of the fertilized eggs and clear eggs.

[A Point Raman Scanning Method for Rapidly Detection of Spinach Quality and Safety Parameters].

According to actual market demand for nondestructive detection of vegetables quality and safety, combined with the heterogeneity of quality and safety parameters such as pesticide residues on leaf vegetables surface and to realize the automatic point scanning for the whole leaf vegetables samples, a suction device based on laboratory (self-designed) Raman spectroscopy hardware and a GUI application software based on the LabVIEW development platform were developed. This system can test the Raman spectroscopy of the whole spinach including the automatic collection, display and storage of the Raman signal of all the scanned points by set up different scan step. A new method to remove the Raman spectrum background was proposed based on data replacement with linear equation at the range of threshold spectrum on both sides of the effective peaks according to the characteristics of spinach original spectra. Its principle is to determine the starting position of linear fitting by judging whether there is trough on both sides of the crest, and then to generate and replace the original spectra data in peak position through the linear fitting equation. Spinach samples were used for the experiment showed that the chlorophyll content and distribution of chlorpyrifos pesticide residue on each scanning point can be obtained after scanning. Therefore, the point scanning Raman system could improve detection accuracy of the quality and safety parameters for the non-uniform samples effectively.

[Detection of Chlorpyrifos on Spinach Based on Surface Enhanced Raman Spectroscopy with Silver Colloids].

In this research, the surface enhanced Raman spectroscopy (SERS) technique is used to develop a nondestructive and fast detecting method for the detection of residual chlorpyrifos on spinach. Silver colloids used for SERS spectroscopy is prepared by the reduction of silver nitrate with hydroxylamine hydrochloride at alkaline pH. The prepared silver colloids are dropped onto spinach samples, then the SERS spectra are collected non-destructively with a self-developed Raman system. This method can be made without physical contact to samples, and rapidly completed without time-consuming sample pre-treatments, and suited to the development of real-time on-line detection methods for trace pesticide residues. SERS signals are collected from 20 points on each spinach sample with 450 mW laser power and 2.5 s exposure time. Chlorpyrifos concentrations in 24 samples are determined with gas chromatography after SERS spectra taken. Savitzky-Golay (SG) smoothing filter and effective peak linear fitting method are used to remove the random noise and the fluorescence background for improving the accuracy of SERS results. The SERS signals are collected from different parts of 50 spinach samples with the same concentration of chlorpyrifos but at different fresh degrees. The relative standard deviation (RSD) of chlorpyrifos' characteristic peak intensities is 13.4%. Although the differences of samples lead to differences in the curves of Raman spectrum, they have little influence on the characteristic peak intensities, which indicates the stability of the proposed detecting method. After the fluorescent background removed, the 20 curves of each sample are averaged. Correlation analysis is done between chlorpyrifos concentration and signal intensity at every Raman shift. Results show that correlation coefficients are higher than 0.85 in the range of 615.5~626.4 cm-1. Signals in this range are used to establish multiple linear regression (MLR) model for the prediction of residual chlorpyrifos. MLR model was developed for chlorpyrifos concentration versus Raman signal intensity at 615.5~626.4 cm-1 for predicting residual chlorpyrifos content in samples, the correlation coefficients of calibration (RC) and validation (RP) are 0.961 and 0.954, which indicate a good linear relationships between them. The minimum detectable threshold for this method is 0.05 mg·kg-1 which is close to the value limited by the national standard of China (0.1 mg·kg-1 for chlorpyrifos in spinach). The proposed practical method is sample, fast, without sample preparation, thus it shows great potential in safety detection of fruits and vegetables.

Determination of Tomato's SSC and TS Based on Diffuse Transmittance Spectroscopy.

In order to meet the demands for rapid and safe nondestructive testing of fruit and vegetable quality，tomato detection system with a special circular light source was built based on the visible / near infrared diffuse transmission principle. Taking soluble solids content (SSC) and total sugar (TS) as the internal quality index, the prediction of 58 tomato samples was carried out by using this system. First, we collected the spectral data of four points for each tomato. Second, Savitzky-Golay smooth(SG-Smooth), standard normal variable transformation(SNV), multiplication scattering correction(MSC), first derivative (FD) and other methods were used to process the original spectral curve before the partial least squares regression(PLSR) model was established. Finally, we validated the established model. The results show that the correlation coefficient (r) of calibration and prediction of the SSC prediction model -are 0.995 6 and 0.976 0 when using 10 point SG-smooth, and the root mean square error of calibration and prediction are 0.052 4% and 0.082 3%. The partial least square regression (PLSR)model, with respect to the first derivative (FD) spectra, provides better prediction performance for total sugar of tomato, with correlation coefficient (r) of calibration of 0.969 1 and 0.972 9, and prediction, root mean standard error of 0.423 8% and 0.454 9%. In the experimental verification of the prediction model, the relationship of SSC between predicted and true value is 0.985 5, root mean square error is 0.066 3°Brix, the relationship of TS between predicted and true value is 0.944 9 while root mean square error is 0.571 5%. The results show that the content of soluble solids and total sugar in tomato can be realized by using visible / near infrared diffuse reflectance spectroscopy. It provides a real-time, nondestructive and rapid detection method for the evaluation of the internal quality of tomato, and provides a theoretical basis for its online grading.

[Simultaneous Detection of Multiple Quality Parameters of Pork Based on Fused Dual Band Spectral].

For dual band visible/near infrared spectroscopy system (350~1 100 and 1 000~2 500 nm), there exsits a band overlap and for the same sample the reflectivity data were unlike due to the performance difference between instruments. A band connection and data fusion method was proposed in this paper to make better use of the dual-band data. A dual-band visible/near-infrared spectroscopy system was built in the study to collect 60 pork samples' reflectance spectra. The reflectance spectra of samples were performed with pretreatment methods of Savitzky-Golay (S-G) and standard normal variable transform to eliminate the spectral noise. Then partial least squares regression (PLSR) prediction models of pork quality attributes (color, pH and cooking loss) based on single-band spectrum and dual-band spectrum were established, respectively. For the cross of two band overlap, the data were connected and integrated using the method put forward in this paper and then PLSR models were established based on the integrated data. The PLSR model yielded prediction result with correlation coefficient of validation (R(p)) of 0.948 8, 0.920 0, 0.950 5, 0.930 1 and 0.903 5 for L(*), a(*), b(*), pH value and cooking loss, respectively. To simplify the model, uninformative variables elimination (UVE) was employed to select characteristic variables. The experimental results show that the proposed method was able to achieve a better fusion of the two band spectral data, and it was good for the establishment of a more simplified and better prediction model.

[Research on Identification and Determination of Pesticides in Apples Using Raman Spectroscopy].

Raman spectroscopy combined with chemometric methods has been thought to an efficient method for identification and determination of pesticide residues in fruits and vegetables. In the present research, a rapid and nondestructive method was proposed and testified based on self-developed Raman system for the identification and determination of deltamethrin and acetamiprid remaining in apple. The peaks of Raman spectra at 574 and 843 cm(-1) can be used to identify deltamethrin and acetamiprid, respectively, the characteristic peaks of deltamethrin and acetamiprid were still visible when the concentrations of the two pesticides were 0.78 and 0.15 mg · kg(-1) in apples samples, respectively. Calibration models of pesticide content were developed by partial least square (PLS) algorithm with different spectra pretreatment methods (Savitzky-Golay smoothing, first derivative transformation, second derivative transformation, baseline calibration, standard normal variable transformation). The baseline calibration methods by 8th order polynomial fitting gave the best results. For deltamethrin, the obtained prediction coefficient (Rp) value from PLS model for the results of prediction and gas chromatography measurement was 0.94; and the root mean square error of prediction (RMSEP) was 0.55 mg · kg(-1). The values of Rp and RMSEP were respective 0.85 and 0.12 mg · kg(-1) for acetamiprid. According to the detect performance, applying Raman technology in the nondestructive determination of pesticide residuals in apples is feasible. In consideration of that it needs no pretreatment before spectra collection and causes no damage to sample, this technology can be used in detection department, fruit and vegetable processing enterprises, supermarket, and vegetable market. The result of this research is promising for development of industrially feasible technology for rapid, nondestructive and real time detection of different types of pesticide with its concentration in apples. This supplies a rapid nondestructive and environmentally friendly way for the determination of fruit and vegetable quality and safety.

[A method for assessing the total viable count of fresh meat based on hyperspectral scattering technique].

The objective of this study is to develop a hyperspectral imaging system to predict the bacteria total viable count in fresh pork. The hyperspectral scattering data were curvefitted by different fitting methods, and correlation differences of models were compared based on the bacteria total viable count of fresh pork, thus providing modeling basis of device for future study. Total 63 fresh pork samples which was used in the experiment were stored at 4 degrees C in the refrigerator of constant temperature. Experiment was performed everyday for 15 days. 4 or 5 random samples were used each day for the experiment. Hyperspectral scattering images and spectral scattering optical data in the wavelength region of 400 to 1 100 nm were acquired from the surface of all of the pork samples. Lorentz and Gompertz function and the modified function was applied to fit the scattering profiles of pork samples. Different parameters could be obtained by Lorentz and Gompertz fitting and the modified function fitting. The different parameters could represent the optical characteristic of the scattering profiles. The standard values of the bacteria total viable count of pork were obtained by classical microbiological plating methods. Because the standard value of the bacteria total viable count was big, log10 of the bacteria total viable count obtained by classical microbiological plating was used to simplify the calculation. Both individual parameters and integrated parameters were explored to develop the models. The multi-linear regression statistical approach was used to establish the models for predicting pork the bacteria total viable count. Both Lorentz and Gompertz function and the modified function included three and four parameters formula. The results showed that correlation coefficient of the models is higher with Lorentz three parameters combination, Lorentz four parameters combination and Gompertz four parameters combination than the individual parameters and other two or three integrated parameters. The three models' correction set and prediction set correlation coefficients were 0.93, 0.96, 0.96 and 0.90, 0.90, 0.92, and the corresponding standard deviation were 0.47, 0.44, 0.39 and 0.56, 0.46, 0.42. Correlation was best with Gompertz four parameters. The device system will select the best correlation and the best stability of the model as the final model.

[Near-infrared hyperspectral imaging combined with CARS algorithm to quantitatively determine soluble solids content in "Ya" pear].

The present study proposed competitive adaptive reweighted sampling (CARS) algorithm to be used to select the key variables from near-infrared hyperspectral imaging data of "Ya" pear. The performance of the developed model was evaluated in terms of the coefficient of determination(r2), and the root mean square error of prediction (RMSEP) and the ratio (RPD) of standard deviation of the validation set to standard error of prediction were used to evaluate the performance of proposed model in the prediction process. The selected key variables were used to build the PLS model, called CARS-PLS model. Comparing results obtained from CARS-PLS model and results obtained from full spectra PLS, it was found that the better results (r(2)pre = 0. 908 2, RMSEP=0. 312 0 and RPD=3. 300 5) were obtained by CARS-PLS model based on only 15. 6% information of full spectra. Moreover, performance of CARS-PLS model was also compared with PLS models built by using variables got by Monte Carlo-uninformative variable elimination (MC-UVE) and genetic algorithms (GA) method. The result found that CARS variable selection algorithm not only can remove the uninformative variables in spectra, but also can reduce the collinear variables from informative variables. Therefore, this method can be used to select the key variables of near-infrared hyperspectral imaging data. This study showed that near-infrared hyperspectral imaging technology combined with CARS-PLS model can quantitatively predict the soluble solids content (SSC) in "Ya" pear. The results presented from this study can provide a reference for predicting other fruits quality by using the near-infrared hyperspectral imaging.

[Rapid nondestructive detection of water content in fresh pork based on spectroscopy technique combined with support vector machine].

Visible near infrared reflectance spectra in the range of 350 nm to 1700 nm were collected from 98 pork samples to develop online, rapid and nondestructive detection system for water content in fresh pork Median smoothing filter (M-filter), multiplication scatter correlation (MSC) and first derivative (FD) were used as compound preprocessing method to reduce noise present in the original spectrum. Seventy four samples were randomly selected to develop training model and remaining 24 samples were used to test the model. The optimal punishment parameters for the support vector machine (SVM) were determined by using cross--validation and grid--search in the training set. SVM prediction model was developed with the radial basis function (RBF) and the developed model was compared with the model developed by partial least squares regression (PLSR) method. SVM prediction model based on RBF had the correlation coefficient and root mean standard error of 0.96 and 0.32 respectively in the training set. The model obtained correlation coefficient of 0.87 and root mean square error of 0.67 in the test set. The result thus obtained demonstrates the applicability of SVM model for rapid, nondestructive detection of water content in pork.

[Rapid nondestructive detection of apple quality attributes using hyperspectral scattering images].

The research discussed the prediction method of apple's internal quality such as firmness and soluble solids content with the combination of parameters getting from hyperspectral fitting scattering curve. The research compared different molding methods using the combination of the three Lorentzian fitting parameters with partial least squares (PLS), stepwise multiple linear regression (SMLR) and neural network (NN). The normalized combination parameters and original combination parameters were used to establish prediction models, respectively. The partial least squares (PLS) prediction models using the combination of three original parameters gave a better results with the correlation of calibration Rc = 0.93, the standard error of calibration SEC = 0.56, the correlation of validation R = 0.84, and the standard error of validation SEV = 0.94 for firmness of apples. The partial least squares (PLS) prediction models using combination of normalized parameters also gave a good results with Rc = 0.95, and the standard error of calibration SEC= 0. 29, the correlation of validation Rv = 0. 83, and the standard error of validation SEV = 0.63 for soluble solids content of apples. The research showed that using hyperspectral scattering curve can detect apple quality attributes at the same time.

[Measurement of chlorophyll content in wheat leaves using hyperspectral scanning].

The objective of the present research was to evaluate the potential of hyperspectral scanning as a way for nondestructive measurement of chlorophyll content in wheat leaves, which can indicates the plant healthy status. One hundred twenty samples were randomly picked from Xiao Tangshan farm. Ninety samples were used as calibration set and others were used for verification set. After capturing hyperspectral image in the range of 400-1,000 nm, the chlorophyll contents of samples were measured immediately. Four different mathematical treatments were used in spectra processing in the wavelength range of 491-887 nm: multiplicative scatter correction (MSC), first derivative correction, and second derivative correction. Statistical models were developed using partial least square regression (PLSR), and stepwise multiple linear regression (SMLR) analysis technique. The results showed that the best calibration model was obtained by PLSR analysis, after processing spectra with MSC and second derivate, with a relatively higher coefficient of determination of calibration (0.82) and validation (0.79) respectively, a relatively lower RMSEC value (0.69), and a small difference between RMSEC (0.69) and RMSEP (0.71). The results indicate that it is feasible to use hyperspectral scanning technique for nondestructive measurement of chlorophyll content in wheat leaves.

[Study of spatially resolved hyperspectral scattering images for assessing beef quality characteristics].

Hyperspectral scattering techniques were used to predict beef pH, tenderness (i. e. WBSF: Warner-Bratzler Shear Force) and color parameters. Thirty-three fresh strip loin cuts were collected from 2-day postmortem carcass. After capturing scattering images and measuring pH values, the samples were vacuum packaged and aged to seventh day, then their color parameters (L*, a*, b*) and WBSF were measured as references. The optical scattering profiles were extracted from the hyperspectral images and fitted to the Lorentzian distribution (LD) function with three parameters. LD parameters, such as the peak height, full scattering width at half maximum (FWHM) and the scattering asymptotic were calculated at individual wavelength. Stepwise regression was used to determine optimal combinations of wavelengths for each of parameters. The optimal combinations were then used to establish multi-linear regression (MLR) models to predict the beef attributes. The full cross validation method was used to examine the performance of models. The models were able to predict beef WBSF with R(CV) = 0.86, and with the SE(CV) (the standard error of cross validation) of 11.7 N, 91% classification accuracy could be obtained. Two-day pH values with R(CV) = 0.86, SE(CV) = 0.07 and color parameters (L*, a*, b*) with R(CV) of 0.92, 0.90 and 0.88, with the SE(CV) of 0.90, 1.34 and 0.41 were obtained respectively. This research provided available technique for the development of multispectral system, which could be implemented online to determine beef steaks color and tenderness.

[Study on modeling method of total viable count of fresh pork meat based on hyperspectral imaging system].

Once the total viable count (TVC) of bacteria in fresh pork meat exceeds a certain number, it will become pathogenic bacteria. The present paper is to explore the feasibility of hyperspectral imaging technology combined with relevant modeling method for the prediction of TVC in fresh pork meat. For the certain kind of problem that has remarkable nonlinear characteristic and contains few samples, as well as the problem that has large amount of data used to express the information of spectrum and space dimension, it is crucial to choose a logical modeling method in order to achieve good prediction result. Based on the comparative result of partial least-squares regression (PLSR), artificial neural networks (ANNs) and least square support vector machines (LS-SVM), the authors found that the PLSR method was helpless for nonlinear regression problem, and the ANNs method couldn't get approving prediction result for few samples problem, however the prediction models based on LS-SVM can give attention to the little training error and the favorable generalization ability as soon as possible, and can make them well synchronously. Therefore LS-SVM was adopted as the modeling method to predict the TVC of pork meat. Then the TVC prediction model was constructed using all the 512 wavelength data acquired by the hyperspectral imaging system. The determination coefficient between the TVC obtained with the standard plate count for bacterial colonies method and the LS-SVM prediction result was 0.987 2 and 0.942 6 for the samples of calibration set and prediction set respectively, also the root mean square error of calibration (RMSEC) and the root mean square error of prediction (RMSEP) was 0.207 1 and 0.217 6 individually, and the result was considerably better than that of MLR, PLSR and ANNs method. This research demonstrates that using the hyperspectral imaging system coupled with the LS-SVM modeling method is a valid means for quick and nondestructive determination of TVC of pork meat.

[A rapid nondestructive measurement method for assessing the total plate count on chilled pork surface].

The present paper proposed a method based on the hyperspectral technology for rapidly, nondestructively quantify the total plate count on chilled pork surface. In the research, 50 chilled pork samples stored at 4 degrees C for 1-14 days were used to study the relationship between the total plate count on chilled pork surface and their hyperspectral images collected in 400-1 100 nm. Two models were established using MLR and PLSR methods, and the prediction showed that they can both give satisfactory results with R(v) = 0.886 and 0.863 respectively. The overall research demonstrates that the hyperspectral technology can well quantify the total plate count on chilled pork surface, and so indicates that it is a valid tool to assess the quality and safety properties of chilled pork in the future.