Near-Field Nondestructive Measurements of Complex Permittivity for Woody Breast Detection in Chicken Meat

The expanded availability of low-cost vector networks analyzers has changed the economics of microwave sensing techniques, allowing for use in non-traditional areas such as food science. For example, poultry producers struggle to identify myopathies in chicken breast that compromise the meat's texture. The current industry standard test is to manually feel each chicken breast by hand to test texture. This method has the disadvantages of being both labor intensive and subjective, and an automated system to scan for defects in chicken breast is desired. This paper details our advances in adapting a commercial dielectric sensor, the epsilon measurement probe, into a tool for inline detection of defects in chicken breast. The sensor operates in a similar manner as an open-ended coaxial line but uses a computational electromagnetic inversion method for corelating its signal to dielectric properties of the test medium. Using this method, we demonstrate that the measured properties are sufficient to classify the grade of the chicken breast as measured by expert graders. The ability to detect defects requires a combination of accurate reflection phase/amplitude, low enough frequency to penetrate into the chicken meat, and a wide enough bandwidth to resolve relevant dispersion features in the chicken permittivity. In addition, we demonstrate how the probe can overcome challenges related to in-line processing, such as alignment of specimens or imperfect contact with the sensor.