Two-dimensional correlation spectroscopy combined with deep learning method and HPLC method to identify the storage duration of porcini

Li Wang,Jie-qing Li,Tao Li,Hong-gao Liu,Yuan-zhong Wang

DOI: https://doi.org/10.1016/j.microc.2021.106670

IF: 5.304

2021-11-01

Microchemical Journal

Abstract:Quick and accurate identification of the storage duration of food is essential to food safety. Herein, two methods reported for analyzing the storage duration of porcini were two-dimensional correlation spectroscopy (2DCOS) combined with deep learning method and HPLC method. 2DCOS combined with deep learning improves the resolution of one-dimensional FT-NIR spectrum and enhances the ability to extract features to obtain a high accuracy model (100%). Used HPLC to detect the changes of chemical components in porcini, the content of uridine and adenosine increased with the increase of storage duration, and the content of guanosine did not change significantly. The accuracy of the PLS-DA model test set based on the HPLC data matrix was 55.56–80%. The results show that 2DCOS combined with deep learning is an effective tool to identify the storage duration of porcini. In addition, by analyzing the influence of uridine and adenosine content changes on the 2DCOS image, we found that the intensity of the autocorrelation peak at 5000 cm−1 can reflect the trend of content changes. It shows that 2DCOS has the potential ability to predict the changing trend of uridine and adenosine content. The 2DCOS combined with deep learning method may promote the non-destructive rapid detection and identification of the storage duration of edible mushrooms without complicated samples or data preprocessing.

chemistry, analytical

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to quickly and accurately identify the storage time of porcini to ensure food safety. Two methods were used in the study to analyze the storage time of porcini: two - dimensional correlation spectroscopy (2DCOS) combined with deep - learning methods and high - performance liquid chromatography (HPLC). Specifically, the paper explored the following points:

1. **Improving spectral resolution and feature extraction ability**: Through the method of 2DCOS combined with deep - learning, the resolution of the traditional one - dimensional Fourier transform near - infrared spectroscopy (FT - NIR) was improved, and the ability to extract features from spectral data was enhanced, thus establishing a high - precision model (with an accuracy rate reaching 100%).

2. **Detection of chemical composition changes**: HPLC was used to detect the chemical composition changes of porcini at different storage times. It was found that the contents of uridine and adenosine increased with the increase of storage time, while the content of guanosine did not change significantly. The accuracy rate of the PLS - DA model test set based on the HPLC data matrix was 55.56% - 80%.

3. **Correlation analysis between 2DCOS images and chemical composition changes**: By analyzing the influence of the changes in uridine and adenosine contents on 2DCOS images, it was found that the intensity of the autocorrelation peak at 5000 cm⁻¹ could reflect the change trend of these component contents. This indicates that 2DCOS has the potential to predict the change trend of uridine and adenosine contents.

In conclusion, this paper aims to develop a non - destructive and rapid detection method for identifying the storage time of edible mushrooms, especially porcini, to promote the development of food safety detection technology.