Rapid Identification of Different Roasting Degrees of Cyperus Rhizome Based on Multi-Source Data Fusion and Artificial Intelligence Algorithms

Cyperus Rhizome (CR) has the effect of soothing the liver. The medicinal effects of CR are enhanced after vinegar processing. But the current processing methods for CR are not standardized, strict quality control during the processing of CR is essential. This study employed computer vision technology to extract spectral and texture features from CR samples. It was found that RGB features in the 680-710 nm wavelength range of cross-sections, Local Binary Pattern (LBP) features from both the cross-sections and epidermis, and Gray-Level Co-occurrence Matrix (GLCM) features from the cross-sections, contributed significantly to distinguishing the samples. The Flash gas chromatography electronic nose (Flash GC e-nose) technology successfully identified 19 volatile components, and nine chemical markers for odor differences in CR with different roasting degrees were determined with VIP > 1. The volatile oil content of CR changes after vinegar frying. Therefore, high-performance liquid chromatography (HPLC) was used to determine the content of α-Cyperone and Cyperenone in CR with different roasting degrees, along with the total volatile oil yield, and comprehensive analysis was conducted by combining the extractable content. A comprehensive analysis was conducted by integrating multi-source data and employing multivariate statistical analysis combined with the Whale Optimization Algorithm-Random Forest (WOA-RF) classification model. This model achieved an accurate classification of vinegar-processed CR samples at different roasting levels, with a classification accuracy of 100%. This study provides a novel method and approach for the rapid differentiation and quality control of vinegar-processed CR, contributing to its quality assurance in medicinal applications.

Keywords: Cyperus Rhizome, Roasting degree, Computer vision, Flash GC e-nose, Multivariate statistics, WOA-RF classification algorithms

Suggested Citation: Suggested Citation

Nie, Xinru and Mei, Xi and Zhang, Jiuba and Shen, Ke and Li, Ming-xuan and Chen, Peng and Wu, Yi and Jin, Huaijie and Wu, Zheyuan and Su, Lian-lin and Mao, Chunqin and tulin, lu, Rapid Identification of Different Roasting Degrees of Cyperus Rhizome Based on Multi-Source Data Fusion and Artificial Intelligence Algorithms. Available at SSRN: https://ssrn.com/abstract=5193542 or http://dx.doi.org/10.2139/ssrn.5193542