黄曲霉菌生长发育和毒素累积的高光谱成像检测模型

Aflatoxins have been classified as the most potent carcinogens by the International Agency for Research on Cancer(IARC), which seriously threatens health of humans and animals. It is critical to develop a non-destructive detecting/screening method to separate moldy/aflatoxin-contaminated from clean (healthy) kernels to preserve quality of cereal grains and oil seeds in China prior to storage. To date, there is no report that has comprehensively examined Aspergillus flavus growth and development-toxin accumulation-kernel damage interaction by using hyperspectral imaging (HSI) technique. In the past five years, the applicant has utilized HSI to study the critical aspects of maize-A. flavus-toxin accumulation that has established a solid foundation to ensure the success of the proposed research. The A.flavus-maize kernel model system will be used to develop a new multi-variable data analysis method to examine interactive spectral and image characters, as well as co-characterization of physical and chemical features. A series of comparative experiments will be performed systematically to obtain spatial and temporal optical signatures of growth and development between non-toxigenic/toxigenic A.flavus on artificial medium / maize kernels with different hardness. The optical data will also be confirmed by scanning electron microscopy and fluorescent protein labeling of fungal invasion process, which will then be used to extract specific fingerprint information of fungal infection and toxin biosynthesis. The comprehensive data on optical characterswill be used for developing an optical model to detect and separate moldy/toxin-contaminated kernel from the healthy kernels. The HSI model on dynamic optical fingerprints in A. flavus-aflatoxin-maize kernels has significant broad impact on food safety by preserving quality in other grain and oil crops. Furthermore, the HSI model will also provide new insights on non-destructively determine optical properties of other kernel–colonizing phytopathogens, such as ear rot and food degradation microbes.

谷物中的黄曲霉毒素已被国际癌症研究机构划定为1类致癌物，严重威胁人畜生命健康。探索非破坏性检测霉变/含毒籽粒的方法对保障我国粮油和饲料安全具有重要意义。至今还没有使用高光谱成像(HSI)技术对黄曲霉生长发育-毒素累积-籽粒损坏互作进行系统研究的报道。申请人在过去5年开创性的利用HSI对玉米真菌污染的关键环节进行了研究，为项目实施奠定了扎实基础。本项目利用黄曲霉-玉米籽粒为样本系统开发新的图谱交互、物理化学协同表征的多变量数据分析方法。实验通过对照法获取产毒/非产毒黄曲霉菌株分别在培养基/不同硬度型籽粒上生长发育的时间序列光谱，辅之以荧光蛋白标记真菌侵入籽粒和扫描电镜组织分析，提取真菌感染和毒素合成特异图谱指纹信息，构建霉变/含毒籽粒鉴别模型和毒素定量检测模型。该模型为玉米和其他粮油作物批量快速分检设备开发奠定理论基础，同时也为其他穗腐病原菌和霉变微生物的非破坏性光学特征研究提供理论依据。

谷物中的黄曲霉毒素已被国际癌症研究机构划定为1类致癌物，严重威胁人畜生命健康。探索非破坏性检测霉变/含毒籽粒的方法对保障我国粮油和饲料安全具有重要意义。.本项目主要以黄曲霉菌-玉米/花生籽粒系统为研究对象，试验通过产毒/非产毒黄曲霉菌从培养基上的理想生长环境逐步过渡到不同预处理/不同硬度籽粒的复杂基质依次推进研究。采用高光谱成像(HSI)技术获取等时间序列光谱数据，通过图谱交互、物理化学协同表征的多变量数据分析方法，研究了霉菌生长发育的光学特性及黄曲霉毒素累积指纹提取方法，构建了霉菌生长检测和黄曲霉毒素累积的定量检测模型，可实现霉菌生长、霉菌种类、霉变籽粒的判别以及黄曲霉毒素含量的预测。同时，以红外显微成像技术结合扫描电镜/透射电镜代替原计划中荧光蛋白标记技术，对霉菌侵入籽粒造成的籽粒营养衰耗与组织结构变化进行微观探测追踪，以宏-微观结合、物理化学协同表征的方式，探究了籽粒霉变过程中与黄曲霉菌的互作机制，验证了黄曲霉菌生长与产毒过程中籽粒光谱信号的特征归属。.项目执行期间，搭建了以高光谱成像系统为核心的检测系统装置1套、正在搭建中一套。建立了分别基于S-G平滑、SNV、MSC、1st D和2nd D预处理方法的PLSDA、SVM等多种生长时间判别模型和PLSR、SVR等多种毒素预测模型以及基于Band Math、PCA、MNF等的多种数据分析算法。共发表SCI论文16篇，EI论文9篇，核心论文1篇，其他论文1篇；授权专利5项；获软件著作权15项；培养硕/博士研究生7名。本项目为玉米和其他粮油作物批量快速分检设备开发奠定理论基础，同时也为其他病原菌和霉变微生物的非破坏性光学特征研究提供了理论依据。