Study On Spectral Imaging Characteristics And Rapid Detection Method Of Pork Meat Organic Macromolecules Preserved At Low Temperature

Chilled and frozen storage are the two most important low-temperature preservation methods to preserve fresh pork meat.In order to ensure food quality and safety,it is important for the meat industry to monitor the quality of pork meat during low-temperature storage.As a new rapid and non-destructive detection method,hyperspectral imaging has a good application prospect for on-line monitoring the quality of low-temperature preserved pork meat.During chilled and frozen storage,the most pronounced changes of pork meat are the degradation,oxidation,and denaturation of macromolecules.In this study,the spectral change characteristics of pork meat organic macromolecules and its rapid detection methods by the visible-near infrared（400-1000 nm）and near infrared（1000-2200 nm）hyperspectral imaging system were investigated.In order to improve detection accuracy,the segmented spectral angle（SAM）algorithm,heterospectral two-dimensional correlation spectroscopy（H2D-CS）,and genetic algorithm-partial least squares（GA-PLS）was creatively introduced to extract the feature wavelengths and spectral shape information.And the spectral changes were investigated by the general two-dimensional correlation spectroscopy（G2D-CS）.The main research contents and conclusions are as follows:（1）The protein and adenosine triphosphate（ATP）degradation degree of pork meat during cold storage at 4 ^oC was investigated by the visible-near infrared（400-1000 nm）hyperspectral imaging system.The biogenic amine index（BAI）and K value were used as the indicator of the degradation degree of protein and ATP,respectively.The successive projection algorithm（SPA）and regression coefficient（RC）methods were used to identify the feature wavelengths of the BAI value.The regression coefficient in prediction of the calibration model for predicting BAI value reached 0.957 with root mean square error in prediction（RMSEP）of 4.866 mg/kg.SPA and gray level co-occurrence matrix（GLCM）were used to select the feature wavelengths of K value and extract the textural information of the image at the feature wavelengths,respectively.When the spectral and textural data were integrated for predicting the K value in pork meat（R²P=0.924,RMSEP=4.0%）.An improvement of 14.9%and 21.6%in model performance was obtained compared with to those when spectral or textural data were used alone.The results of G2D-CS analysis showed that the autopeaks obtained when the BAI and K value were used as perturbations were similar during cold storage of pork meat.This indicates that the trend of chemical changes in pork meat during cold storage as the BAI and K value increasing are mostly the same.（2）The rapid detection of myofibrils cold structural deformation degree of frozen pork meat samples obtained from different freezing rates was investigated by the near-infrared hyperspectral imaging system.The myofibrils cold structural deformation degrees were evaluated by surface hydrophobicity（S0ANS）and Ca²⁺-ATP activity.Segmented SAM algorithm was introduced to extract the feature spectral shape information and used for establishing the partial least square regression（PLSR）model.The results showed that the hyperspectral imaging system can be used to extract the near-infrared spectra of the frozen meat without the interference of the frost on the surface of the frozen meat samples,and the segmented SAM algorithm has comparable or even better performance in extracting the spectral feature information(S₀ANS:R_P²=0.896,RMSEP=1.549;Ca²⁺-ATPase:R_P²=0.879,RMSEP=0.015μmol Pi/mg protein/min).As the S0ANS value increasing,the N-H bond changes firstly,then the C-H bond,and finally the O-H bond.（3）The protein oxidation and secondary structure changes during frozen storage of pork were investigated by the near-infrared hyperspectral imaging system.The carbonyl content andα-helix fraction were used as the indicator of protein oxidation and protein secondary structure changes,respectively.The H2D-CS technique was used to identify the carbonyl content andα-helix fraction related wavebands by correlate the near-infrared spectra with the mid-infrared and circular dichroism spectra,respectively.The results showed that the H2D-CS method has a better performance in selecting the carbonyl andα-helix content-related wavelengths（carbonyl content:R²P=0.896,RMSEP=0.177 nmol/mg;α-helix fraction:R²P=0.832,RMSEP=1.824%）compared with the commonly used RC and SPA methods,indicating that the H2D-CS method is an effective feature extraction method that can be applied to the near-infrared spectrum of frozen pork.During the frozen storage of pork,as the carbonyl content increases,the C=O bonds change first,then the N-H bond,and finally the C-H bond.As theα-helix fraction decreasing,the spectral intensity of N-H stretching changed before that of the C-H and O-H,and the spectral fluctuation of the O-H stretching from intramolecular O-H bonds occurred after the carbonyl related C-H band and before those aliphatic and aromatic C-H band.（4）Lipid oxidation degree monitoring of pork meat during frozen storage by near-infrared hyperspectral imaging.The TBARS value was used as the indicator of lipid oxidation degree.The results of MSC preprocessing showed that there was a significant correlation between the TBARS value and ice crystal during frozen storage of pork meat.Based on the calibration models established by the MSC correction coefficients（αandβ）,the TBARS values had almost the same prediction performance（R²P=0.926,RMSEP=0.036）as that established by the full wavebands,and this result was not affected by the frozen storage temperature and time.As the TBARS value increasing,the spectral intensity of C=O stretching changed before that of the C-H.（5）Protein degradation extent traceability of the raw meat used for two kinds of processed meats by visible and near-infrared hyperspectral imaging.The TVB-N value was used as the indicator of protein degradation extent,and GA-PLS was introduced to identify the feature wavelengths.The results showed that the protein degradation related chemical information can still be reflected in the visible and near-infrared spectra after the salting and cooking treatment.And the regression models based on GA-PLS identified feature wavelengths achieved an R²P of 0.882 and RMSEP of 2.231 mg/100g for salted meat,and an R²P of 0.853 and RMSEP of 2.428 mg/100g for cooked meat.After the raw pork meat with different protein degradation degrees are salted,the position of the autopeaks obtained were almost the same or similar with that of the fresh meat,as the protein degradation extent increasing,the spectral intensity changes following the sequence of 624 nm-469 nm-541nm-581 nm.However,after cooking,the number of autopeaks with the same position as that of the fresh meat is reduced.