TL;DR
This paper presents a novel application of Near-Infrared (NIR) spectroscopy combined with chemometrics (specifically Partial Least Squares Regression - PLSR) for the rapid and non-destructive quantification of Total Flavonoids Content (TFC) and Total Antioxidant Activity (TAA) in Ginkgo biloba leaves. The developed method offers a significant improvement over traditional, time-consuming, and destructive analytical techniques, paving the way for efficient quality control of Ginkgo biloba products.
What Problem Does This Paper Address?
Ginkgo biloba leaves are widely recognized for their health benefits, primarily attributed to their flavonoid content and antioxidant properties. However, accurately and efficiently quantifying these key components is crucial for ensuring the quality and efficacy of Ginkgo biloba products. Traditional methods for determining TFC and TAA, such as HPLC and colorimetric assays, are often time-consuming, require extensive sample preparation (extraction), and can be destructive to the sample. This presents a bottleneck for high-throughput quality control in the pharmaceutical and nutraceutical industries. The paper addresses this gap by exploring the potential of NIR spectroscopy, a technique known for its speed and non-destructive nature, to provide a rapid and accurate alternative for quantifying these critical quality indicators.
Methods
The core of this research lies in the development of a robust calibration model using Near-Infrared (NIR) spectroscopy and chemometric techniques. The process involved several key steps:
- Sample Collection and Preparation: 113 batches of Ginkgo biloba leaves were collected, dried, powdered, and sieved to a uniform particle size (60-mesh).
- Reference Analysis: Traditional methods were used to determine the ground truth for TFC (colorimetric assay using rutin as a standard) and TAA (DPPH radical scavenging activity assay using Trolox as a standard).
- NIR Spectra Acquisition: NIR spectra (4000-8000 cm⁻¹) were collected from each sample using a diffuse reflectance mode.
- Data Preprocessing and Outlier Detection:
- Principle Component Analysis (PCA): Used initially to visualize sample distribution and identify potential outliers.
- Mahalanobis Distance and Hat Matrix: These statistical methods were employed to rigorously identify and remove anomalous spectra that could negatively impact model calibration. Four samples were identified as outliers and removed.
- Calibration Model Development (PLSR):
- Sample Splitting: The remaining 108 samples were split into an 80% calibration set (88 samples) and a 20% prediction set (20 samples) using the Kennard-Stone (K-S) algorithm.
- Spectral Preprocessing: Various methods, including Standard Normal Variate (SNV), multiplicative scattering correction (MSC), and Savitzky-Golay derivatives, were evaluated to optimize spectral data. SNV was found to be the most effective.
- PLSR Factor Optimization: A 10-fold cross-validation was performed to determine the optimal number of PLS factors, minimizing the Root Mean Square Error of Cross-Validation (RMSECV).
- Model Validation: The performance of the final PLSR model was assessed using the prediction set, evaluating metrics like the coefficient of determination (R²) and Root Mean Square Error of Prediction (RMSEP).
Datasets
The study utilized a dataset of 113 batches of Ginkgo biloba leaves collected from Zhejiang Province, China. This dataset was divided into a calibration set of 88 samples and a prediction set of 20 samples after outlier removal. No new benchmark datasets were introduced; rather, existing samples were used for both reference analysis and NIR spectral acquisition.
Key Insights and Ideas
- NIR as a Viable Alternative: The study strongly demonstrates that NIR spectroscopy, when coupled with appropriate chemometric methods, can serve as a powerful and rapid tool for quantifying complex phytochemicals like flavonoids and their associated activities in botanical samples.
- Importance of Preprocessing and Outlier Detection: The effectiveness of the NIR calibration model was highly dependent on rigorous data preprocessing, particularly outlier removal using Mahalanobis distance and hat matrix methods, and spectral pretreatment techniques like SNV. This highlights the critical role of data quality in chemometric modeling.
- Simultaneous Quantification: The developed method allows for the simultaneous determination of both TFC and TAA, offering a significant advantage over sequential traditional analyses.
- Non-Destructive Analysis: The non-destructive nature of NIR spectroscopy is a major benefit, allowing for repeated analysis of the same sample and preserving valuable material for other studies.
Key Results
The study successfully developed and validated PLSR models for the quantitative analysis of TFC and TAA in Ginkgo biloba leaves.
- TFC Prediction: The final model achieved a coefficient of determination (R²) of 0.8863 and a Root Mean Square Error of Prediction (RMSEP) of 2.203 mg/g.
- TAA Prediction: The model for TAA achieved an R² of 0.8486 and an RMSEP of 0.2211 mM/g.
These results indicate a high degree of accuracy and reliability in predicting both TFC and TAA using the NIR method.
Limitations and Future Work
- Limitations:
- The study focused on a specific geographical origin of Ginkgo biloba leaves (Zhejiang Province). The transferability of the model to leaves from different regions or with varying cultivation practices might require further investigation.
- While the paper mentions the importance of spectral preprocessing, the detailed optimization process for all tested methods is not exhaustively presented, focusing primarily on the chosen SNV.
- Future Work:
- The authors suggest that this NIR method, combined with chemometrics, has the potential for high sample throughput analysis and low costs, making it suitable for widespread application in the quality control of Ginkgo biloba products.
- Further research could explore the application of this method to other Ginkgo biloba derived products (e.g., extracts, tablets) and investigate the model's robustness across different geographical origins and processing conditions.
Summary updated