Chemical Pattern Recognition and Color-Chromaticity Correlation Analysis for Quality Control of Stir-Fried Perillae Fructus.

Objective: Perillae Fructus (PF) (Perilla frutescens (L.) Britt.) and stir-fried Perillae Fructus (SFPF) are commonly used clinically for the treatment of cough and asthma, yet their quality control methods have not been fully established. Method: The best processing techniques of PF were optimized by one-variable-at-a-time (OVAT) analysis and Box-Behnken design (BBD); fingerprint combined with chemical pattern recognition techniques was employed to establish chromatographic fingerprints of PF and SFPF from different regions. Differential compounds were screened and the reliability of the established method was verified through quantitative analysis of multi-components; image processing technology was applied to determine chromaticity values and perform cluster heatmap analysis. The composition-color correlation of PF and SFPF was investigated. Result: Four characteristic components were identified through 36 batches of PF and SFPF, with rosmarinic acid, 5-hydroxymethylfurfural, caffeic acid and luteolin serving as discriminant markers differentiating PF and SFPF. The contents of seven components and the corresponding chromaticity parameters (L*, a*, b*) were determined to generate a visualized heatmap. Rosmarinic acid and caffeic acid showed positive correlations with L*, whereas a negative correlation was shown with b* and 5-hydroxymethylfurfural. Conclusions: This study provides a theoretical basis for judgment of processing endpoints and the rapid online quality monitoring of SFPF.