Nutritional Estimation of Foods with Hyperspectral Imaging and Machine Learning

Original Title: Hyperspectral imaging as a non-destructive technique for estimating the nutritional value of food

Marín-Méndez et al. Volume 9, 2024, 100799 https://doi.org/10.1016/j.crfs.2024.100799

Knowing the energy and nutrient content of processed foods is essential for both the food industry and public health. Traditional methods are time-consuming, require chemicals, and destroy the sample. This study proposes an alternative: using hyperspectral imaging (NIR and VIS-NIR) combined with mathematical models.

A total of 118 food samples were analyzed using this technology. Based on their spectral data, ten machine learning models were tested to predict key nutrients. Ridge regression showed the highest accuracy, especially for protein (R² = 0.88), moisture (R² = 0.85), and energy (R² = 0.76).

This non-destructive approach enables quick, reliable nutritional analysis, opening up new possibilities for food assessment.

Would you like to know more? Check it out Current Research in Food Science