Estimation of cotyledon isoflavone abundance by a grey luminance-based model in variously hilum-coloured soybean varieties

BACKGROUND: The nutraceutical uses of soybean (Glycine max L. Merr.) have received increasing attention in recent years, due to the therapeutic effects of high seed isoflavone concentrations against heart disease, cancer and menopausal symptoms. RESULTS: We found a close correlation between seed isoflavone abundance and hilum colour in a set of 17 contrasting soybean varieties. Image analysis of the hilum greylevel pattern allowed us to identify a power model which approximates total cotyledon isoflavone concentrations (TCIC) at 65–71% by the normalised modal grey level. Higher TCIC levels were assigned to darker hilum varieties and vice versa within a variety-dependent response. Optimisation of the algorithm required correction for a few specific varieties falling in the intermediate 1.1–1.5 mg/g TCIC range, which were over-estimated by the model, perhaps due to variations in hilar optical properties related to the geometric features of both hilum and seed. CONCLUSION: In view of its easy, low-cost detection, seed hilum colour is a useful phenotypic trait in soybean for rapid evaluation of isoflavone abundance in food uses and for improving specific nutraceutical breeding programmes.