RÉSUMÉ
Objective:To establish a new fast and accurate method for identifying the authenticity and specifications of Fritillariae Cirrhosae Bulbus based on electronic nose technology， and to discuss the feasibility of this technology in the identification of decoction pieces. Method:Fritillariae Cirrhosae Bulbus was used as the research object， 80 batches of samples to be tested were collected， and the olfactory sensory data of the electronic nose were taken as independent variables （<italic>X</italic>）， the results of the method contained in the 2020 edition of <italic>Chinese Pharmacopoeia</italic> were taken as the focus， and the traditional empirical identification results were used as benchmarking information （<italic>Y</italic>）. Four chemometric methods， including discriminant analysis （DA）， least square support vector machine （LS-SVM）， principal component analysis-DA （PCA-DA） and partial least squares-DA （PLS-DA）， were used to establish the identification model ［<italic>Y</italic>=<italic>F</italic>（<italic>X</italic>）］ of authenticity and commodity specifications of Fritillariae Cirrhosae Bulbus， respectively. Wherein， the identification accuracy and time-consuming was taken as indicators to discuss the results. Result:After cross-verification by leave-one-out method， the correct rates of the above four models were 93.75%， 91.25%， 95.00% and 95.00%， respectively， and the PCA-DA and PLS-DA identification models were the best in terms of authenticity identification. In specification identification， the correct rates of these four models were 86.67%， 88.00%， 89.33% and 68.00%， respectively， and the PCA-DA identification model was the best. The electronic nose had a high accuracy in the identification of authenticity and specification model， and the time consuming was relatively short. Conclusion:Electronic nose technology can identify Fritillariae Cirrhosae Bulbus accurately and quickly， and has significant advantages in terms of timeliness and correct judgment rate.
RÉSUMÉ
The quality of traditional Chinese medicine tablets is correlated with clinical efficacy and drug safety, and plays a great role in promoting the development of traditional Chinese medicine. However, the existing traditional artificial identification and modern instrument detection in terms of accuracy and timeliness have both advantages and disadvantages. Therefore, how to quickly and accurately identify the quality of traditional Chinese medicine tablets has become a high-profile issue. The purpose of this paper is to explore the feasibility of the application of electronic eye technology in the study of rapid identification of traditional Chinese medicine quality. A total of 80 batches of samples were collected and tested by Fritillariae Cirrhosae Bulbus for traditional empirical identification(M_1) and modern pharmacopeia(M_2). The optical data was collected from electronic eyes, and the chemical metrology was used to establish suitable discrimination models(M_3). Four authenticity and commodity specification models, namely identification analysis(DA), minimum bidirectional support vector machine(LS-SVM), partial minimum two-multiplier analysis(PLS-DA), main component analysis identification analysis(PCA-DA), were established, respectively. The accuracies of the authenticity identification models were 82.5%, 90.0%, 96.2% and 93.8%, while the accuracies of the commodity specification identification models were 89.3%, 96.0%, 90.7% and 97.3%, respectively. The models were well judged, the authenticity identification was based on the final identification model of PLS-DA, and the commodity specification was based on the final identification model of PCA-DA. There was no significant difference between its accuracy and M_1, and the time of determination was much shorter than M_2(P<0.01). Therefore, electronic-eye technology could be used for the rapid identification of the quality of Fritillariae Cirrhosae Bulbus.