Assessing different processed meats for adulterants using visible-near-infrared spectroscopy

Area of Science:

• Food Science and Technology
• Analytical Chemistry
• Spectroscopy

Background:

• Food adulteration poses significant risks to consumer health and market integrity.
• Rapid and accurate detection methods are crucial for ensuring the quality of meat products.

Purpose of the Study:

• To investigate the efficacy of visible/near-infrared (Vis-NIR) and near-infrared (NIR) spectroscopy for detecting plant and animal protein adulterants in minced beef and pork.
• To apply machine learning techniques for classification, prediction, and wavelength selection in adulterant analysis.

Main Methods:

• Utilized spectroscopic systems in the 400-1000nm (Vis-NIR) and 900-1700nm (NIR) ranges.
• Employed multiple machine learning algorithms for classification (presence/absence, type) and prediction of adulterant levels.
• Evaluated models using selected wavelengths versus full spectral ranges.

Main Results:

• Selected wavelength models demonstrated superior classification and prediction performance compared to full wavelength models.
• Achieved high classification rates: 96% for pure samples and 100% for adulterated samples in the first stage.
• Second stage classification rates ranged from 69-100%, with optimal models for predicting adulterant levels showing correlation coefficients (r) of 0.78-0.86.

Conclusions:

• Spectroscopic technology offers a promising avenue for the rapid and accurate detection of adulterants in minced meat.
• This approach can enhance food safety and quality control measures in the meat industry.