Assessment of Intramuscular Fat Quality in Pork Using Hyperspectral Imaging

Abstract

Intramuscular fat (IMF) quality in muscle contributes to important aspects of meat quality and is critical to the nutritional, sensory values, and shelf stability of meat. The measurement of IMF quality currently relies on gas chromatography (GC), which is labor-intensive and time-consuming. This study investigated the use of hyperspectral imaging to predict the quality of IMF in pork loin cuts. Pork loin cuts were scanned using hyperspectral imaging (900–1700 nm), followed by GC analysis of the fatty acid profile. Mean spectral features (MSF), Gabor filter features (GFF), and wild line detector feature (WLDF) techniques were used to extract texture features and then related with the GC results by partial least-squares regression (PLSR) algorithm. Simplified models were developed using PLSR from selected wavelengths, and the results of the validation provided by the WLD features showed coefficient of determination (R2) for C14:0, C16:0, C16:1, C18:0, C18:1, C18:2, SFA (saturated fatty acid), MUFA (monounsaturated fatty acids), and PUFA (polyunsaturated fatty acid) that ranged from 0.805 to 0.942 and root mean square error of prediction ranged from 0.087 to 0.304 mg/g meat. The result indicates that texture features from hyperspectral images could be used to develop a rapid tool for assessment of the IMF quality in the pork.