• Suivez-nous

Discrimination of the class origin of bones present in the sediment fraction of animal by-products using NIRM.


  • De la Haba, M. , Fumière, O. , Garrido Varo, A. , Guerrero, J. , Perez Marin, D. , Dardenne, P. , Baeten, V. & Fernández Pierna, J.A. (2007). Discrimination of the class origin of bones present in the sediment fraction of animal by-products using NIRM. G.R. Burling-Claridge, S.E. Holroyd and R.M.W. Sumner. Proceedings in: 12th International Conference on Near Infrared Spectroscopy (ICNIRS): NIR in action, Auckland - New Zealand, 11-15/04/2005, New Zealand Near Infrared Spectroscopy Society Incorporated, 95-98.
Type Conference Proceedings
Year of conference 2007
Title Discrimination of the class origin of bones present in the sediment fraction of animal by-products using NIRM.
Editor G.R. Burling-Claridge, S.E. Holroyd and R.M.W. Sumner.
Conference name 12th International Conference on Near Infrared Spectroscopy (ICNIRS): NIR in action
Conference location Auckland - New Zealand
Publisher New Zealand Near Infrared Spectroscopy Society Incorporated
Recnumber 282
Pages 95-98
Label U15-0500
conference Date 11-15/04/2005
Endnote Keywords animal by-products|near infrared microscopy (NIRM)|discrimination|class origin|Support Vector Machines (SVM)|
Abstract Discrimination of the class origin of bones present in the sediment fraction of animal by-products using Near Infrared Microscopy (NIRM) de La Haba M.J. 1, Fernández J.A. 2, Fumière O. 3, Garrido A. 1, Guerrero J.E. 1, Pérez-Marin, M.D, 1 Dardenne P. 3 and Baeten V. 3 1 Faculty of Agriculture and Forestry Engineering. ETSIAM. University of Córdoba. Apdo. 3048. 14080 Córdoba. Spain 2 Scientific collaborator F.N.R.S. Statistics and Informatics Department, University of Agronomical Sciences, Avenue de la Faculté 8, 5030 Gembloux, Belgium. 3 Walloon Agricultural Research Centre (CRA-W), Quality of Agricultural Products Department, Chaussée de Namur n°24, 5030 Gembloux, Belgium. (baeten@cra.wallonie.be) According to the European Legislation, there is an urgent need to develop reliable, fast and accurate forensic methods that could identify at class level (e.g. terrestrial vs. fish) and detect at low level (i.e. 0.1 %) animal ingredients in raw materials and compound feeds. Classical microscopy is currently the only official method available to identify ingredients of vegetal and animal origin in a mixture. However, this method is time consuming and laborious. Its performance depends largely on the experience of the analyst and is mainly based on the presence of constituents easily identifiable (i.e. bone fragments). Moreover, its use for the discrimination of class origin of the animal by-products detected in a compound feed is limited: it only allows discrimination between terrestrial bones and fish bones. Recent studies have demonstrated that the performance of the detection of meat and bone meal (MBM) from terrestrial origin at low level (i.e. 0.1%) drops drastically if fish meal is also present in the compound feed to be tested. Recently, Near Infrared Microscopy (NIRM) has been successfully used for the detection, at low level, of animal particles in the sediment fraction of compound feeds. The detection of MBM is then based on the use of the infrared spectra of the particles. In this study, the potential of NIRM to discriminate the class origin of the animal particles included in the sediment fraction of animal by-products (ABP) has been investigated. In this way, a spectral library was built with well identified ABP samples from different species ( cattle, ovine, pig and poultry) provided by the European Fats Processors and Rendering Association (EFPRA) and fish meal samples provided by a Belgian rendering plant and the International Federation for Fish and Oils (IFFO). For each sample, a total of 30 particles were scanned (30 spectra/sample). Principal Component Analysis (PCA) and the Mahalanobis distance (H) were performed to identify and to eliminate possible outliers included in the spectral libraries. Different chemometric classification techniques (PLS, ANN and SVM) have been tested in order to construct discriminant models able to classify the animal particles according to their origin. The results show that it is possible to use NIRM to discriminate the particles according to their terrestrial or fish origin with a rate of success higher than 90 %. Moreover, further work is in progress in order to produce models able to discriminate the particles according to their mammalian or avian origin.
Author address Baeten Vincent, Quality Department of Agro-food Products, Walloon Agricultural Research Centre (CRA-W), Chaussée de Namur, 24, B-5030 Gembloux, baeten@cra.wallonie.be
Fichier Discrimination of the class origin of bones present in the sediment fraction of animal by-products using NIRM.
Caption U15-0500-fernandez-2005.pdf
Authors De la Haba, M. , Fumière, O. , Garrido Varo, A. , Guerrero, J. , Perez Marin, D. , Dardenne, P. , Baeten, V. & Fernández Pierna, J.A.