Fresh Meat Classification Using Laser-Induced Breakdown Spectroscopy Assisted by LightGBM and Optuna

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Machine Learning

Background:

• Accurate identification of fresh meat varieties is crucial for food safety and quality control.
• Traditional methods for meat classification can be time-consuming and labor-intensive.
• Laser-induced breakdown spectroscopy (LIBS) offers a rapid, non-destructive approach for elemental and molecular analysis.

Purpose of the Study:

• To develop and validate a highly accurate method for classifying fresh meat varieties (pork, beef, chicken) using LIBS.
• To compare the performance of the LightGBM model with support vector machine (SVM) for LIBS spectral data analysis.
• To optimize the LightGBM model hyperparameters using the Optuna algorithm for enhanced classification accuracy.

Main Methods:

• Fresh meat tissue samples (pork, beef, chicken) were analyzed using LIBS after surface flattening.
• Spectral data (900 spectra) were collected from plasma generated on meat surfaces.
• Feature selection was performed using information gain and peak extraction algorithms.
• The LightGBM model was optimized with Optuna, and the SVM model was optimized using 10-fold cross-validation.

Main Results:

• The optimized LightGBM model achieved a classification accuracy of 0.9370, macro-F1 of 0.9364, and Cohen's kappa coefficient of 0.9244.
• The SVM model achieved lower performance metrics: 0.8888 accuracy, 0.8881 macro-F1, and 0.8666 kappa coefficient.
• The LightGBM model demonstrated superior performance in distinguishing between fresh meat varieties.

Conclusions:

• The combined approach of LIBS and the optimized LightGBM model provides a novel and effective method for rapid fresh meat classification.
• This technique has the potential to significantly improve the efficiency and accuracy of meat inspection processes.
• Further research can explore the application of this method to a wider range of meat products and potential contaminants.