Microscopic method in processed animal proteins identification in feed: applications of image analysis

• Main Authors: Savoini G, Dell'Orto V., Cheli F., Tognon G., Campagnoli A., Pinotti L.
• Format: Article
• Language: English
• Published: Presses Agronomiques de Gembloux 2004-01-01
• Series: Biotechnologie, Agronomie, Société et Environnement
• Subjects: Processed animal proteins (PAP); official microscopic method; image analysis
• Online Access: http://www.pressesagro.be/base/text/v8n4/249.pdf

Processed animal proteins (PAP) detection and identification in feedstuffs can be difficult in distinguishing among land animals, i.e. poultry and mammals. Thus, the aim of this study was to evaluate the potential application of image analysis in PAP identification. For this purpose four reference samples containing poultry meals and four reference samples containing mammalian meat and bone meals were used. Each sample was analyzed using the microscopic method (98/88/EC). Bone fragments are characterized by similar morphological features (colours, shape, lacunae shape, lacunae distribution, etc.) that make it diff i c u l t to distinguish between poultry and mammals. Through a digital camera and an image analysis software a total of 30 bone fragment lacunae images at X400 were obtained. For each image 29 geometric parameters related to the lacunae and 3 geometric parameters related to the canaliculae of lacunae, were measured using the image analysis software obtaining 960 observations. Of the 32 descriptors used two, the area of the lacunae and their perimeter, were able to explain 96.15% of the total variability of the data, even though their contribution was different (83.97% vs. 12.18%, respectively). Through these two descriptors it was possible to distinguish between mammalian and poultry lacunae, except in two cases (6.6%), in which poultry lacunae were wrongly classified as mammalian. This latter can be related with higher variability in the lacunae area recorded for mammals compared to poultry. On the basis of the present study, it can be concluded that image analysis represents a promising potential tool in PAP identification, that may provide accurate and reliable results in feedstuffs characterisation, analysis and control.