[Producing area identification of Letinus edodes using mid-infrared spectroscopy]

Area of Science:

• Agricultural Science
• Analytical Chemistry
• Spectroscopy

Background:

• Accurate identification of agricultural product origin is crucial for quality control and consumer trust.
• Traditional methods for determining the geographical origin of food products can be time-consuming and labor-intensive.
• Spectroscopic techniques offer a rapid and non-destructive approach for analyzing food composition and origin.

Purpose of the Study:

• To investigate the efficacy of mid-infrared (MIR) spectroscopy for identifying the producing area of Letinus edodes (shiitake mushrooms).
• To develop and compare classification models using various machine learning algorithms for origin identification.
• To evaluate the performance of relevance vector machine (RVM) as a novel classification technique in this context.

Main Methods:

• MIR spectra of Letinus edodes were acquired and preprocessed using Multiplicative Scatter Correction (MSC).
• Classification models were built using five techniques: Partial Least Squares-Discriminant Analysis (PLS-DA), Soft Independent Modeling of Class Analogy (SIMCA), K-Nearest Neighbor Algorithm (KNN), Support Vector Machine (SVM), and Relevance Vector Machine (RVM).
• Effective wave numbers were selected using weighted regression coefficients (Bw) from the PLS-DA model to optimize classification performance.

Main Results:

• All developed classification models achieved over 80% accuracy.
• KNN, SVM, and RVM models, utilizing full spectra, demonstrated excellent performance with over 90% classification accuracy in both calibration and prediction sets.
• RVM models consistently performed well, achieving over 90% classification accuracy regardless of whether full spectra or selected effective wave numbers were used.

Conclusions:

• Mid-infrared spectroscopy is a viable and effective tool for identifying the producing area of Letinus edodes.
• The Relevance Vector Machine (RVM) algorithm shows significant promise as a novel and high-performing classification technique for spectroscopic analysis.
• This study provides a new conceptual framework for quality analysis of Letinus edodes and other agricultural products using MIR spectroscopy.