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

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.