Rapid Determination Of Freshness Changes In Tan Mutton During Refrigerated Storage By Hyperspectral Imaging

Tan sheep,a dominant and characteristic livestock breed in Ningxia,its meat products are tender and have a uniform distribution of fat,which are favored by consumers.For meat products,freshness,as one of the most important quality attributes,has a strong relationship with product sales and consumption.Conventionally,sensory evaluation and chemical analysis are used to evaluate meat freshness,which have some shortcomings,such as strong subjective judgment,complex sample pretreatment and tedious determination steps.Therefore,some simple,sensitive,economical and reliable analytical methods with real-time evaluation function are needed to achieve rapid and non-destructive detection of freshness in Tan mutton.In this paper,the content of TBARS,TBA and IMP and its degradation compounds were used as the evaluation indexes of meat freshness.This study investigated the feasibility of hyperspectral imaging technique coupled with chemometrics analysis and generalized two-dimensional correlation spectra(G2D-CS)for the detection and evaluation of Tan mutton freshness.The change rule between freshness changes and spectra imaging of Tan mutton was elucidated to provide a theoretical reference for realizing intelligent monitoring and fast non-destructive analysis of Tan mutton quality.The main research conclusions were as follows:(1)Rapid determination of TBARS contents in Tan mutton using hyperspectral imaging.The spectra reflectance information of Tan mutton were collected using near-infrared hyperspectral imaging system.The TBARS contents in Tan mutton were determined by spectrophotometry.Spectra reflectance and chemical value of the sample were correlated.The raw spectra were corrected by utilizing mathematical pretreatment methods and reduced dimensionally by variable selection algorithm.The PLSR?RF and LSSVM prediction models of TBARS contents were established based on key variables.The result showed that the calibration model based on OSC pre-processing yielded acceptable results,with a Rc of 0.94,Rp of 0.84 and RMSEP of 0.19 mg/kg.The OSC-CARS-LSSVM was the optimal spectra detection model for predicting TBARS content(Rc=0.91,Rp=0.91,RPD=2.81).At the same time,G2D-CS technique was used to analyze the sequence change in the NIR spectra of mutton with TBARS contents as the perturbation.The spectral changes caused by the fluctuation of TBARS content occurred following the sequence of 1443 nm-1140 nm-1666 nm-1005 nm.This result indicated that as the content of TBARS increased,the spectral fluctuation of N-H antisymmetric stretch 1st overtone(1443 nm)occurred before that of the N-H second overtone of muscle proteins(1140 nm),and the spectral fluctuation of the first overtone of C-H stretching(1666 nm)occurred after the N-H second overtone and before that of 2nd overtone of O-H stretching(1005 nm).(2)Rapid determination of biogenic amine contents in Tan mutton using hyperspectral imaging.The changes of TBA contents in Tan mutton during refrigerated storage were investigated.The TBA contents were regarded as external disturbance,and the dynamic changes of spectral signals were studied by G2D-CS analysis to find the sensitive areas associated with TBA contents.The PLSR linear and LSSVM nonlinear model were established to evaluate the effectiveness of feature variables extracted by various selection algorithms.The result showed that with the increase of the degree of mutton samples corruption,the levels of TYR,CAD and PUT significantly increased.Compared with the raw spectra,the PLSR model established by MF pretreatment showed the highest correlation coefficient for evaluating TBA contents,with Rp of 0.88,RMSEP of 2.18 mg/kg and RPD of 2.11.According to G2D-CS analysis,it could be summarized that the spectral changes caused by the fluctuation of TBA contents occurred following the sequence of 1002 nm-1204 um-1335 nm.The wavelengths in the spectra range of 1002-1335 nm were selected as the detection area for TBA contents in Tan mutton.iVISSA,VCPA and CARS were used to select key wavelengths from G2D-CS analysis to develop simplified models.The G2D-CS-CARS-LSSVM model had good robustness and prediction accuracy(Rc=0.92,Rp=0.91,RPD=2.76).(3)Rapid determination of the content of IMP and its degradation compounds in Tan mutton using hyperspectral imaging.The spectra reflection information of Tan mutton were collected to build the prediction model of the content of IMP and its degradation compounds.The best model was identified by comparing the modeling effects of 5 preprocessing methods and different variable selection algorithm.The result showed that the full spectra model of IMP had the best predictive effect after MF preprocessing(Rc=0.92,Rp=0.87,RPD=2.26).The best predictive model of HxR was NormalizePLSR(Rc=0.93,Rp=0.90,RPD=2.33).The performance of Normalize-PLSR model was better than other models for evaluating Hx contents(Rc=0.94,Rp=0.89,RPD=2.43).The characteristic variables of spectra data were extracted by iVISSA,VCPA and CARS algorithm.The MF-CARS-LSSVM model showed excellent results for assessing IMP contents(Rc=0.90,Rp=0.95,RPD=3.95).For parameter HxR,the Normalize-CARS-LSSVM model had the best predictive effect(Rc=0.92,Rp=0.98,RPD=4.76).The Normalize-CARS-LSSVM model of Hx contents had the best robustness and predictive ability(Rc=0.94,Rp=0.96,RPD=4.29).The spectral changes caused by the fluctuation of IMP contents occurred following the sequence of 1008 nm-1145 nm-1320 nm based on G2D-CS analysis.(4)Visualization distribution of freshness chemical information in Tan mutton during cold storage.The freshness chemical information corresponding to each pixel in the spectra image were predicted using the optimized PLSR analysis model.The predicted values were displayed on the mask image to realize the visualization expression of TBARS,biogenic amines,IMP and its degradation compounds.By comparing and analyzing the visualization distribution map of freshness chemical information,the dynamic changes of Tan mutton freshness could be understood clearly.