Counterfeit Citri Reticulatae Pericarpium Identification: Multi-Class Detection Using Vis/NIR Spectroscopy and...

Area of Science:

• Pharmacognosy
• Analytical Chemistry
• Artificial Intelligence

Background:

• Citri Reticulatae Pericarpium (CRP) faces significant counterfeiting due to high market demand and value.
• Accurate, non-destructive methods are crucial for identifying counterfeit CRP.

Purpose of the Study:

• To develop and evaluate a novel detection model for authenticating Citri Reticulatae Pericarpium (CRP).
• To compare the performance of deep learning models for counterfeit discrimination.

Main Methods:

• Visible/near-infrared (Vis/NIR) spectroscopy was used to collect spectral data from authentic and counterfeit CRP samples.
• Wasserstein Generative Adversarial Network with Gradient Penalty (WGAN-GP) was employed for data augmentation.
• Deep learning models, including Inception-ResNet, were applied for classification and detection.

Main Results:

• The Inception-ResNet model achieved 96.56% accuracy in identifying authentic CRP from four counterfeit types.
• The proposed deep learning approach outperformed conventional machine learning methods.
• WGAN-GP data augmentation improved model generalization and classification robustness.

Conclusions:

• The integrated Vis/NIR spectroscopy and deep learning model offers a precise and efficient solution for non-destructive authentication of CRP.
• This method effectively addresses the challenge of adulteration analysis in valuable botanical materials.
• The study highlights the potential of advanced AI techniques in quality control for traditional medicines.