Non-destructive Detection and Screening of Non-uniformity in Microwave Sterilization Using Hyperspectral Imaging Analysis

The non-uniformity of microwave sterilization usually causes incomplete inactivation of microorganism (especially foodborne pathogens) and potential food safety issue. Detecting the non-uniformity of load distribution using hyperspectral imaging (HSI) (400-1050 nm) in tandem with multivariate analysis was investigated and infrared (IR) thermal imaging technique was used to assist the evaluation. The best simplified model of least-squares support vector machines (LS-SVM) obtained based on ten optimal wavelengths showed an R-p(2) of 0.905, RMSEP of 0.404 log CFU/mL and RPD of 2.62, while the analysis of temperatures detected by IR thermal camera showed that the temperatures of mid-way along each edge were obviously higher than the center, and heating time of microwave had a significant effect on mean temperature (p < 0.01) and temperature coefficient of variation (COV) values (p < 0.01). Moreover, the visualization of L. monocytogenes load distribution was acquired by transferring the quantitative model of LS-SVM to each pixel in the image at different heating times, with the results being well consistent with those of temperature distribution by IR thermal camera. Therefore HSI technique could be used to monitor the non-uniformity of microwave sterilization processing.