J.A. Fernández Pierna

Walloon Agricultural Research center (CRA-W), Gembloux, Belgium
Invited by Laura Symul

will give a presentation on :

Applications of Vibrational Spectroscopy and Chemometrics in the Agro-Food Sector

Abstract:

Vibrational spectroscopy, as Near infrared (NIR) or Raman, is the most widely used non-destructive technology in the food and feed industries for the daily determination and quantification of qualitative parameters of the materials. The high throughput of the method, the capacity to determine in one single analysis a panoply of parameters, the possibility to build a network of spectrometers together with its potential use both on-line and at-line in a production plant made this technique even more attractive. These techniques provide real-time analyses with an increased sample throughput. Moreover, more recent areas as hyperspectral imaging allows collection of spectroscopic images at different levels from single kernel or particle levels to satellite. This is of great interest for laboratories that control feed compound or cereals. Other decisive advantages of spectroscopic methods are the ability to determine simultaneously different parameters and criteria, no use of reagents and reduced sample preparation.

The combination of these techniques with appropriate data treatment, chemometrics or machine learning tools should help to solve the deep and rapid changes that the agro-food sector is facing with increasing consumer concerns about food and feed safety and quality issues. Chemometrics and machine learning are increasingly applied to vibrational spectroscopic data to enhance data analysis and interpretation. Chemometrics methods like principal component analysis (PCA) and partial least squares (PLS) are commonly used to extract key features from complex spectra, while machine learning techniques such as support vector machines (SVM), random forests, and neural networks improve pattern recognition, classification, and quantification. These approaches enable the handling of large datasets, improve the accuracy of detecting chemical or biological properties, and assist in applications like quality control, environmental monitoring, and disease diagnosis.

In summary, there is an increased need for appropriate techniques and methods to help producers, retailers, and processors to control and to track their products. Vibrational spectroscopy combined with chemometrics should allow to build strategies that can be applied to check (on-line, at-line and at the laboratory level) the quality of food and feed materials, to detect non conformity and subsequently to identify targeted or untargeted adulterants and contaminants among others.