Submitted to: ASAE Annual International Meeting
Publication Type: Proceedings
Publication Acceptance Date: August 1, 2002
Publication Date: N/A
Interpretive Summary: All chickens consumed by U.S. consumers are required to be inspected post-mortem by the United States Department of Agriculture/Food Safety and Inspection Service (USDA/FSIS) inspectors for wholesomeness. Current inspection for wholesomeness is based on visual methods. There are six major defects that cause chicken carcasses to be removed from the processing line. They are septicemia, cadaver, bruise, tumor, airsacculitis, and ascites. Under the current Hazard Analysis and Critical Control Point (HACCP) -based Inspection Models Project (HIMP), FSIS requires that any poultry showing evidence of septicemia to be condemned. The Instrumentation and Sensing Laboratory has been actively involved in the development of machine vision systems which can provide an alternative means to inspect poultry carcasses with a high degree of accuracy on a consistent basis. An unwholesome poultry carcass category, septicemia, has been successfully classified with a high accuracy. The current ISL inspection system relies on the use of reflectance techniques. This investigation examined fluorescence spectroscopy as a complementary approach to the reflectance method and further incorporated additional poultry disease categories. With the use of fluorescence techniques, four chicken categories, normal, airsacculitis, cadaver, and septicemia, were classified with 98% accuracy. These research results are critical to scientists (e.g., engineers) who are developing automated machine vision poultry inspection systems. This is also beneficial to the poultry processing industry.
Fluorescence spectroscopy was used to characterize chicken carcass spectra. Spectral signatures of three different disease categories of poultry carcasses (airsacculitis, cadaver, and septicemia) were obtained from fluorescence emission measurements in the wavelength range of 360 to 600 nm with 330 nm excitation. Principal Component Analysis (PCA) was used to select the most significant wavelengths for the classification of poultry carcasses. These wavelengths were analyzed for pathologic correlation of poultry diseases. Using a Soft Independent Modeling of Class Analogy (SIMCA) of principal components with a Mahalanobis distance metric, poultry carcasses were individually classified into different classes with 97.9 percent accuracy.