Hybrid method for accurate starch estimation in adulterated turmeric using Vis-NIR spectroscopy

Area of Science:

• Food Science
• Analytical Chemistry
• Spectroscopy

Background:

• Turmeric is a valuable commodity in South East Asia, used as a food additive and health supplement.
• Its economic importance makes turmeric susceptible to adulteration with cheaper substances like starch.
• Rapid and accurate detection methods are crucial to ensure turmeric purity and safety.

Purpose of the Study:

• To develop and validate a spectroscopic method for quantifying starch adulteration in turmeric.
• To assess the effectiveness of various spectral preprocessing and feature selection techniques.
• To evaluate the performance of stacked generalization models for accurate adulterant detection.

Main Methods:

• Reflectance spectra of turmeric-starch mixtures (0-50% starch) were collected (400-2050 nm).
• Spectral data underwent preprocessing including Multiplicative Scatter Correction (MSC) and Standard Normal Variate (SNV), followed by Savitzky-Golay (SG) filtering.
• Feature selection utilized Extra Tree Regressor (ETR), dimensionality reduction via Principal Component Analysis (PCA), and stacked generalization with K-Nearest Neighbours (KNN), Decision Tree (DT), Random Forest (RF), Logistic Regression (LRC), and Linear Regression (LR) models.

Main Results:

• The proposed stacked generalization model achieved excellent regression performance (r²=0.999, RMSE=0.206, RPD=73.73, RER=480.58).
• Classification achieved 100% accuracy, indicated by F1 score and Matthew's Correlation Coefficient (MCC).
• The method demonstrated high sensitivity and accuracy in detecting and quantifying starch adulteration.

Conclusions:

• Spectroscopic analysis combined with advanced chemometric methods offers a robust solution for detecting turmeric adulteration.
• The developed method provides a reliable, rapid, and accurate tool for quality control in the food industry.
• Ensuring turmeric authenticity protects consumers and upholds market integrity.