基于表面增强拉曼散射技术的复杂体系中贝类重金属污染的快速检测及机理研究

Shellfish has strong capacity of heavy metal accumulation. The modernization progress has caused a series of heavy metal pollution. And thus it is the urgent need for a rapid detection method to ensure the safety of shellfish food. Raman spectroscopy technology, featured with simple, real-time, free from water interference et al, has been well known as one of the fastest growing food safety inspecting technology. In this work, four typical heavy metal ions including copper, zinc, lead and cadmium, will be used to manually culture the shellfish Tegillarca granosa. The clathrate form and chemical structure of the target constituent, which was different between the control and contaminated samples, will be carefully analyzed the heavy metals connected. The nanospheres pointer will be designed purposely according to the weak signal of selected oriented-enhanced Raman site, and synthesized with functional multi-quantum-point sites. These sites will be attached particularly to the functional groups of target sites by means of pH, temperature regulation and ultrasonic pretreatment et al. Interactive adsorption of between nanosphere and target proteins, and mechanism of surface-enhanced Raman scattering will be investigated. Due to the complex and unknown information in the actual process, the conventional closing-set model can’t predict the known information beyond the model’s “trained knowledge”, thus the proposed opening-set detection model will be studied to solve the ‘rejected’ unknown information. Given that can’t be enumerated as many as possible the pollution classes or samples, the semi-supervised one-class classification will be introduced to expand the information capacity, aiming at identifying the healthy shellfish. This project can enrich the approaches to non-invasively detect heavy metals in shellfish, and has high significance to ensure the quality and safety of shellfish food in China.

贝类富集重金属能力强，受现代化进程带来一系列重金属污染问题，亟需一种快速检测方法来保证贝类食品安全。拉曼光谱技术具有简便、实时、不受水分干扰等特点，被誉为发展最快的食品安全检测技术之一。项目以贝类泥蚶为研究对象，以铜、锌、铅、镉四种典型重金属离子为污染指标，分析重金属在泥蚶体内的络合形态及化学结构。针对泥蚶组分拉曼信号弱的问题，研究筛选特异性的增强拉曼散射位点，设计、合成自主装功能化的多量子点纳米球探针，并辅以酸碱度、温度调节、超声波等处理，探讨纳米球与目标位点的吸附以及对表面增强拉曼散射信号的作用机制。针对复杂污染体系下未知信息，常规闭集模型无法检测超出模型的已知信息，研究开集检测模型来解决未知信息的误识问题；同时不能枚举尽可能多的污染类样本，研究半监督单类分类器来扩张信息容量，识别健康类样本。本项目可以丰富贝类重金属快速检测手段，保障我国贝类食品的质量安全具有重要意义。

贝类富集重金属能力强，受现代化进程带来一系列重金属污染问题，亟需一种快速检测方法来保证贝类食品安全。研究以浙江温州地区养殖的贝类泥蚶为研究对象，人工饲养正常情况下（即健康样本）以及不同重金属胁迫下（包括铜、锌、镉、铅以及4种重金属联合胁迫）的泥蚶样本。经过一定时间的饲养，使其富集不同浓度的重金属离子。设计多量子点纳米球探针，辅以不同条件与泥蚶肉质进行接触，经拉曼光谱技术分析，纳米球探针难以获取结合态的重金属离子。而后，对泥蚶肉质进行一系列的预处理，包括冷冻干燥、磨粉、压片，采集其拉曼光谱、激光诱导击穿光谱、红外光谱，重点分析了重金属铜离子与光谱之间的定量关系。研究以自组织映射神经网络技术来筛选特征变量，这相比于常规算法，具有更好的结果；另外，研究了共识建模技术来提升复杂环境下模型性能的稳定性，这将充分利用了光谱不同特征信息；对于未知信息的样本预测，采用线性回归稀疏化分类法，是基于最近邻方法和单类目标的线性子空间，确定离预测样本最近的单类目标组成的线性子空间，这可以降低样本的误分类。本项目可以丰富贝类重金属快速检测手段，保障我国贝类食品的质量安全具有重要意义。