羊肉细胞中过氧化物酶的分子高光谱检测及其波谱响应机理研究

Tan-sheep is one of the best advantaged agricultural products which have extremely local characteristics in NingXia. Due to the lack link in slaughter of scientific detection means，the phenomenon of disease flesh sales frequently occurred, which are serious harm to human bodies. The traditional physical and chemical detection method of disease flesh was time-consuming and experiment process trival, which was difficult to meet the requirement of rapid precise detection. Since the rapid detection research of Disease flesh composed a molecule cell level by hyperspectral imaging techniques. The POD content of disease flesh was one of the important indicators that was chosen as the research object and the fresh meat frozen biopsy samples was scanned by molecular hyperspectral system, the total absorption spectra of multi-component would be separated by adopting the methods of PCA, MNF and so on, then the characteristic vector of the POD eigenvectors in cells would be extracted, and combined with image processing technologies, chemical imaging and spatial visualization of single component would be completed. With the help of the migration of macro forecast model, the rapid detection method of cell migration trace material would be established. Based on quickly identification of POD content in the cell and chemical imaging levels, the research framework of quantitative analysis of POD and spatial distribution of visualization would be established, and the characteristic wavelength spectral interference factors of specific compound extracted and wavelength migration effect would be explored. The response mechanism of spectral absorption of the groups of certain cells containing hydrogen were revealed and the high accuracy rapid detection technology system of disease flesh would be achieved. The results of the research will provide theoretical basis for the study of rapid detection of lamb disease and residues of veterinary drugs and other microscopic substances.

滩羊肉是宁夏极具地方特色的优势农产品之一。当前由于屠宰环节缺乏科学的检测手段使病害肉上市销售现象频繁发生，对人体危害严重；传统病害肉理化检测方法耗时长，实验过程繁琐，难以满足快速精准检测要求，由此提出采用分子高光谱技术从细胞水平对病害肉进行快速检测研究。过氧化物酶（POD）是病害肉诊断检测的重要指标，试验以羊肉中POD为研究对象，利用分子高光谱采集新鲜羊肉样本显微图谱，采用PCA、MNF等方法实现多组分吸收光谱剥离，提取出细胞内POD特征向量，借助宏观预测模型迁移建立细胞内微区痕量物质快速检测方法，课题从细胞内POD含量快速识别及化学成像两个层面，构建POD定量分析及空间分布可视化研究框架，探究特定化合物提取特征波长的干扰因素及波峰偏移效应，揭示光谱吸收对细胞特定基团的响应机理，构建病害肉特征物高精度原位快速检测技术体系，为羊肉病害特征物及兽药残留等微观物质快速检测技术研发提供理论参考。

病害肉特征物快速识别是保障肉品安全的重要手段。研究基于显微高光谱成像技术，对羊肉肌肉组织中含有的SOD酶活力、POD酶活力、可溶性蛋白含量指标进行检测，结合光谱图像信息与化学指标二者内在相关性，建立模型并进行可视化研究，结合光谱物质基团对特征波段光谱吸收初步揭示光谱吸收对细胞内物质的响应机理，通过物质基团吸收能量后发生伸缩、弯曲、旋转等，解析背景化合物对光谱吸收的干扰因素及波峰偏移效应。试验搭建显微高光谱系统，确定系统扫描参数，对超氧化物歧化酶（SOD）、过氧化物酶（POD）活力值、可溶性蛋白含量进行高光谱数据建模分析。根据模型得出两种酶的高光谱建模效果均为：Micro-NIR 900＞Micro-Vis 400＞NIR 900＞Vis 400高光谱。SOD酶活力的最优模型为经SNV预处理后，用SPA提取特征波长的LSSVM模型，模型共12个波长，Rc=0.8900，Rp=0.8521；POD酶活力的最优模型为SG处理后，经CARS提取特征波长的LSSVM模型，模型共14个波长，Rc=0.8680，Rp=0.8410。对可溶性蛋白含量的四种高光谱数据进行建模分析，得出最优光谱依次为Vis 400＞NIR 900＞Micro-NIR 900＞Micro-Vis 400，最优模型为经过MSC预处理后，用CARS提取特征波长的PLSR模型，模型共13个波长，Rc=0.8702，Rp=0.8063。利用SOD酶活力、POD酶活力和可溶性蛋白含量三个化学指标与高光谱数据结合进行可视化研究。根据化学值指标建立基于最优预处理的PLSR模型，利用β系数法提取光谱输入变量，得到可视化成像效果图。融合特征光谱信息与图像纹理信息实现对油酸、棕榈酸、硬脂酸及亚油酸含量的快速检测及分布可视化研究，结果显示：较最优特征波长数据模型，融合纹理信息后，油酸含量PLSR预测模型R2 c与R2 p分别升高了0.0056与0. 0027；棕榈酸含量LS-SVM预测模型的R2 c与 R2 p分别为0.8859、0.9502，高于单一光谱数据模型；硬脂酸含量LS-SVM模型R2 c升高了0.0722；亚油酸含量PLSR模型R2 c与R2 p分别升高了0.0761与0. 1090，表明融合数据建立的模型其预测性能优于单一光谱信息模型；最后将预测结果反演到样本图像上生成可视化分布图.