Rapid Identification of Lingnan Aromatic Traditional Chinese Medicinal Materials Using Deep Learning-Based Headspace-Gas Chromatography-Ion Mobility

Aromatic herbs are an important part of Lingnan traditional Chinese medicine. However, the quality control of these herbs faces challenges in ensuring authenticity, compliance, and distinguishing traditional processing techniques, etc. The study introduces a novel approach for the rapid identification of aromatic traditional Chinese medicinal materials from the Lingnan region using a deep learning-enhanced headspace-gas chromatography-ion mobility spectrometry (HS-GC-IMS). HS-GC-IMS technology, a method that merges the efficiency of gas chromatography with the high resolution of ion mobility spectrometry, was employed to gather intricate fingerprint data of volatile compounds in traditional Chinese medicinal materials. Further explore the deep meaning of the “odor” fingerprint information of Chinese medicinal materials using deep learning models, aiming for the classification of aromatic Chinese medicinal materials. Experiment results demonstrate that EfficientNet-B0 has superior performance, and outperforming other models like ResNet18 and MobileNetV2.