Development and validation of near infrared microscopy spectral libraries of ingredients in animal feed as a first step to adopting traceability and authenticity as guarantors of food safety

Abstract

Traceability of animal products has become a priority for governments of the developed countries as a guarantee of food safety. Near infrared microscopy (NIRM) has been proposed as an alternative technology to detect and quantify banned ingredients in feedstuffs. The great advantage of this technique is its objectivity, whilst retaining the sensitivity of classic microscopy. The aim of this work was to build an NIRM reference spectral library on animal feed, consisting of samples of animal feed ingredients and possible contaminants, and to assess its ability to discriminate between ingredients using an internal crossvalidation. A total of 48,899 spectra were measured on 229 samples representing 30 different ingredients. The method chosen for classification was K-nearest-neighbours (KNN) using first derivative spectra. Although the results showed an overall classification error of 35.88%, there was good discrimination between ingredients of animal and vegetable origin. There was some confusion between similar vegetable ingredients but this is unimportant.