Detection of Edible Insect Kolbe (Protaetia brevitarsis seulensis) Powder Adulteration Using FTIR Spectroscopy and Chemometrics: A Regression-Based Analysis

¹ Department of Biosystems Engineering, College of Agricultural and Life Sciences, Gyeongsang National University, 501, Jinju-daero, Jinju-si, Gyeongsangnam-do, Republic of Korea
² Department of Biosystems Engineering, College of Agriculture, Life, and Environment Sciences, Chungbuk National University, Cheongju 28644, Republic of Korea
³ Department of Biosystems Machinery Engineering, College of Agriculture and Life Science, Chungnam National University, Daejeon 34134, Republic of Korea
⁴ Department of Smart Agriculture Systems, College of Agriculture and Life Science, Chungnam National University, Daejeon 34134, Republic of Korea
⁵ Department of Agricultural Engineering, Faculty of Food Technology and Agroindustry, Universitas Mataram, Mataram 83115, Indonesia
⁶ Institute of Smart Space Agriculture, Gyeongsang National University, 501, Jinju-daero, Jinju-si, Gyeongsangnam-do, Republic of Korea

^* Corresponding author: geonwookim@gnu.ac.kr

Abstract

Edible insects are an alternative food source rich in protein and minerals and support global food security. However, the lack of mass production and the few farmers who developed it make the product price expensive and widely counterfeited. Fourier-transform infrared (FTIR) is one of the effective methods in detecting product authenticity; a combination using chemometrics models is used to predict chickpea adulteration in kolbe (Protaetia brevitarsis seulensis) powder. The adulteration was made by adding the chickpea flour to the Kolbe powder with various concentrations from 5 % to 50 % of weight basis and followed by FTIR spectra acquisition. Pre-processing methods such as Multiplicative Scatter Correction (MSC) and Savitzky-Golay smoothing are used to improve spectral data quality and reduce noise. Three typical machine learning models, such as partial least squares regression (PLSR), support vector regression (SVR), and XGBoost, were applied to develop a quantitative model. Our study demonstrated that the SVR model with MSC spectra was the best-fit model in estimating the chickpea flour concentration, denoting an R² of 0.979 with an RMSE of 2.508. This study shows that FTIR spectroscopy combined with a chemometric regression model is a powerful non-destructive method for detecting adulteration of edible insect powders in food matrices.