Feasibility study for the surface prediction and mapping of phytonutrients in minimally processed rocket leaves (Diplotaxis tenuifolia) during storage by hyperspectral imaging

•Reliable quantification of Vitamin C in rocket leaves using NIR.•First derivative followed by mean centering served as best pretreatment.•Average vitamin C represented by pixels within calibration range decreased with time.•Detected pixels within calibration range decreased with the passage of storage time. A comprehensive study of the feasibility of hyperspectral imaging in visible (400–1000 nm) and near infrared (900–1700 nm) regions was investigated for prediction and concentration mapping of Vitamin C, ascorbic acid (AA), dehydroascorbic acid (DHAA) and phenols in wild rocket (Diplotaxis tenuifolia) over a storage span of 12 days at 5 °C. Partial least squares regression (PLSR) with different data pretreatments and wavelength selection resulted in satisfactory predictions for all parameters in the NIR range except DHAA. Prediction models were used for concentration mapping to follow changes over time. The prediction maps will be comprehensively study to assess the pixel to pixel variation within the rocket leaves. The PLSR models for Vitamin C, AA and phenols yielded an R2 of 0.76, 0.73 and 0.78, respectively in external prediction with root mean square errors approximately equivalent to those of reference analysis. Conclusively, hyperspectral imaging, with the correct mapping approach, can be a useful tool for the prediction and mapping of phytonutrients in wild rocket (Diplotaxis tenuifolia) over time.