MRI ‘texture’ analysis of MR images of apples during ripening and storage

Abstract

Images obtained by magnetic resonance imaging (MRI) of the apple varieties Idared, Redspur and Topaz during ripening and storage were analysed by texture analysis (TA) to determine the correlation between TA parameters and firmness, soluble solids content (SSC) and titratable acids. All varieties differ significantly in SSC, acidity and the firmness of fruits. Three sagital T2w slices measured at 4.7 T (resolution 256×256 pixels) were analysed numerically using the software MaZda 2.11 and the statistical package S-PLUS 4.5. A significant correlation was found between acidity and TA parameters such as sum average, sum variance and sum entropy. The dynamics of TA parameters during the periods of maturation and storage were described by polynomial functions. The following distinctive TA parameters were found: skewness and kurtosis (among histogram-based parameters), variance of absolute gradient (among gradient-based parameters), grey level nonuniformity (among run length matrix-based parameters) and especially some co-occurrence matrix-derived parameters such as correlation, sum average, sum variance and sum entropy (within 1-, 3-, and 5-pixel neighbourhoods). The TA parameter correlation obtained from MR images can be used for characterising fruit ripening.