Field Tool to Measure Defects in Chicken Breast via Microwave Dielectric Properties

In recent years the poultry industry has focused on detecting and eliminating meat quality disorderssuch as wood-breast white-striping and spaghetti-meat. The current industry method for detectionin a production environment is a manual technician-based scoring method that provides unknownaccuracy and limited speed. We propose a method to detect these meat quality disorders in chickenbreast fillets by a non-contact microwave measurement that can be practically adapted to high- speed poultry production environments. Previous studies indicate that microwave dieletricspectroscopy can determine many features of the internal states of chicken breast. Two separatefeasibility studies were performed where chickens with varying degrees of meat quality werescored by an expert and measured with our microwave probes. The measurements were processedby a basic classification algorithm based on a linear algebra technique known as "Singular ValueDecomposition" which identifies the meat quality disorders with approximately 75% accuracysolely based on output signal with no corrections or physical modeling. We studied chickensamples at varying degrees of processing: from live birds to fully processed fillets.In the proposed work focused on processed chicken breast fillets we will we develop a tool for thedetection of chicken breast quality with microwave probes. We will focus on three directions: 1)developing physics-based models to extract meaningful physical properties (such as protein %water % etc) from the microwave measurements of chicken breasts 2) developing algorithms thataccuractely classify chicken breast quality from raw data and 3) developing a prototype conveyor- belt-based system to take measurements at speed classify chicken quality and display results via atechnician-friendly interface.