CN-121978230-A - GC-MS and hyperspectral imaging fusion-based yam decoction piece origin tracing method and application

Abstract

The invention discloses a method for tracing origin of yam decoction pieces based on GC-MS and hyperspectral imaging fusion and application thereof, comprising the steps of S1 sample preparation, S2 data acquisition, S21 hyperspectral data acquisition, S22GC-MS data acquisition, S3 multisource data characteristic analysis, S4 characteristic screening and data fusion, S5 model construction and performance verification and S6 fusion fingerprint construction. The data fusion model solves the problems of easiness in fitting and poor generalization capability of a single technology, and the tracing result is more reliable. The random forest model based on the fusion data achieves the classification accuracy of 92.3 percent, which is obviously higher than that of a single data source method. The fusion fingerprint provides a unique visual identification for each production place, and is beneficial to technical transformation and industrial popularization.

Inventors

• WANG ZENGLI
• CHENG MUYUN
• Cui Rujia
• WANG YUCHEN

Assignees

• 中国农业大学

Dates

Publication Date

20260505

Application Date

20251214

Claims (10)

1. 1. A method for tracing origin of yam decoction pieces based on GC-MS and hyperspectral imaging fusion is characterized by comprising the following steps: s1, preparing a sample; S2 data acquisition, specifically comprising S21 hyperspectral data acquisition and S22GC-MS data acquisition: S3, multi-source data characteristic analysis; s4, feature screening and data fusion; S5, constructing a model and verifying performance; S6, constructing a fusion fingerprint.
2. 2. The method of claim 1, further comprising S7 mechanism analysis and statistical verification, and/or S8 classification performance verification.
3. 3. The method of claim 1, wherein S1 comprises taking fresh white mouth yams from different places, washing, scraping, peeling, slicing, and drying.
4. 4. The method of claim 1, wherein the hyperspectral data acquisition comprises acquiring hyperspectral images of the yam decoction piece samples using a hyperspectral imaging system. And (3) carrying out black-and-white plate correction on the original image after acquisition, extracting the region of interest of each sample, and obtaining the average reflectivity spectrum curve of each sample under different wavelength channels.
5. 5. The method of claim 1, wherein the GC-MS data acquisition comprises precisely weighing rhizoma Dioscoreae decoction piece powder in a test tube, adding acetonitrile, soaking, vortex oscillating, and ultrasonic extracting. Standing, taking supernatant, adding derivatization reagent, and carrying out ultrasonic derivatization reaction after vortex mixing. After the reaction is finished, cooling to room temperature, performing GC-MS analysis, and comparing with a standard mass spectrum library to identify the compound.
6. 6. The method of claim 1, wherein the multi-source data characterization comprises (A) hyperspectral characterization curves showing significant differences in characteristic wavelengths for different production sites, (B) GC-MS chemical analysis showing significant differences in content distribution of key chemical markers between different production sites, (C) characteristic wavelength distribution box line graphs further verifying spectral differences between production sites, and (D) chemical marker variability heat graphs showing stability of chemical components inside each production site.
7. 7. The method of claim 6, wherein the feature screening comprises (1) hyperspectral feature wavelength screening, wherein the competitive adaptive re-weighting sampling algorithm is used for screening a plurality of feature wavelengths from a hyperspectral full band, and 5 feature wavelengths are 818nm, 756nm, 946nm, 938nm and 742nm. (2) And (3) screening the GC-MS key difference chemical markers, namely screening a plurality of key difference chemical markers, namely citric acid, mannitol, inositol, palmitic acid and fructose, from GC-MS data by adopting partial least squares discriminant analysis and variable importance projection and variance analysis.
8. 8. The method of claim 7, wherein the data fusion comprises the steps of respectively performing Z-score normalization processing on absorbance data of 5 characteristic wavelengths and peak area data of 5 chemical markers, splicing the absorbance data and the peak area data into a 10-dimensional fusion characteristic vector, and constructing a fusion data matrix; The model construction and performance verification comprises the step of dividing fusion feature vectors of large samples and production place labels thereof into training sets and testing sets in proportion. And respectively using a support vector machine and a random forest algorithm to construct a production place discrimination model.
9. 9. The method of claim 2, wherein the fused fingerprint construction comprises constructing a "chemical index-spectral index" joint fingerprint for each origin, (a) the fingerprint in the form of a radar map visually shows unique distribution patterns of yam pieces in 10 characteristic dimensions of different origins, (B) the fingerprint in the form of a heat map quantitatively characterizes standardized distribution of characteristic values of each origin, and provides a quantifiable "identity authentication" identifier for each origin; And/or Mechanism analysis and statistical verification include, demonstrate the distribution characteristic of the fusion data in the classification of the place of origin through principal component analysis.
10. 10. An application of a Chinese yam decoction piece origin tracing method based on GC-MS and hyperspectral imaging fusion in identifying Chinese yam decoction piece origins.

Description

GC-MS and hyperspectral imaging fusion-based yam decoction piece origin tracing method and application Technical Field The invention relates to the technical field of traditional Chinese medicine quality detection and tracing, in particular to a system and a method for tracing the origin of yam decoction pieces by combining hyperspectral imaging and gas chromatography-mass spectrometry (GC-MS) technology. Background Yam is a bulk medicinal material with homology of medicine and food, and the quality of the yam is closely related to the place of production, namely has 'genuine property'. The genuine medicinal materials refer to medicinal materials produced under specific natural conditions and ecological environment, the production is concentrated, and the cultivation technology and the harvesting processing have specific requirements, so that the quality and the curative effect of the same medicinal materials produced in other areas are better. The origin of the yam decoction pieces in the current market is mixed, so that the problems of secondary filling and the like exist, and the clinical medication safety and the consumer rights are seriously influenced. The existing method for identifying the place of origin of the Chinese yam decoction pieces mainly comprises traditional sensory evaluation and modern instrument analysis. The traditional sensory evaluation method mainly depends on the identification of the experienced pharmacist through the modes of eye observation, hand touch, nose smell, mouth taste and the like, and has strong subjectivity, poor reproducibility and difficult quantification. The modern instrument analysis method mainly comprises chromatographic techniques such as gas chromatography-mass spectrometry (GC-MS), high Performance Liquid Chromatography (HPLC) and the like, can accurately identify chemical components in the traditional Chinese medicine, and has the advantages of high sensitivity and high resolution. However, chromatographic techniques are complex to pre-process, time consuming, costly and are a lossy assay, and difficult to use for large-scale rapid screening. Spectroscopic techniques, such as near infrared spectroscopy, hyperspectral imaging techniques, etc., have the advantage of being fast, non-destructive, and field detectable. The hyperspectral imaging technology can simultaneously acquire the spatial information and the spectral information of the sample, and has good application prospect in traditional Chinese medicine identification. However, the identification result of the spectrum technology lacks clear chemical substance indication, the interpretation of the model is poor, and the authority of the model is often questioned. The single technology tracing has obvious limitations that the chromatographic technology can provide accurate chemical component information, but the treatment process is complex and is not suitable for rapid screening, and the spectroscopic technology is rapid and nondestructive, but lacks chemical mechanism support and has limited result reliability. Therefore, a technology for tracing and fusing the origin of the traditional Chinese medicine materials, which can achieve the detection efficiency, the precision and the result reliability, is urgently needed in the field. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a yam decoction piece origin tracing method based on the fusion of GC-MS and hyperspectral imaging, and the rapid, accurate and interpretable identification of the yam decoction piece origin is realized by the fusion of the rapid nondestructive testing capability of hyperspectral imaging and the accurate chemical analysis capability of GC-MS. The method comprises the following steps: s1, preparing a sample; S2 data acquisition, specifically comprising S21 hyperspectral data acquisition and S22GC-MS data acquisition: S3, multi-source data characteristic analysis; s4, feature screening and data fusion; S5, constructing a model and verifying performance; S6, constructing a fusion fingerprint. Further, the method also comprises S7 mechanism analysis and statistical verification and/or S8 classification performance verification. Preferably, S1 comprises taking fresh rhizoma Dioscoreae from different places, cleaning, peeling with scraper, slicing, further drying. For example, the materials are dried for 5 hours at a low temperature (50 ℃) by using an open type drying box, cooled and placed in a dryer for standby. Several samples were prepared for hyperspectral imaging analysis at each place of origin, and several parallel samples were taken for GC-MS analysis at each place of origin. Preferably, the hyperspectral data acquisition comprises the step of acquiring hyperspectral images of the yam decoction piece samples by using a hyperspectral imaging system (the spectral range is 400-1000 nm). And (3) carrying out black-and-white plate correction on the original image after acquisition, extracti