Effect of varying optical properties on the modeling of astaxanthin concentration in salmon by visible spectroscopy

The purpose of this work was to investigate how natural changes in optical properties in salmon muscle during storage affected the color of salmon and the prediction of astaxanthin based on spectroscopic measurements. Atlantic salmon (Salmo salar) were stored for 6 days at 2–4 °C and measured once per day by an interactance probe at 449–744 nm. RGB (Red, Green, Blue) images were taken and were used to calculate average L*a*b* values and redness index (a*/b*). The interactance data revealed a systematic time dependent variation in the absorption at 450 and 600 nm. The redness index increased and the L* value decreased during storage. Partial least squares regression was used to calibrate the interactance spectra from day one against values of astaxanthin determined by high performance liquid chromatography. The calibration model was then used to predict astaxanthin concentrations in the same samples based on spectroscopic measurements performed during the following days. The optical changes were large enough to affect the prediction errors of astaxanthin significantly. A storage time of six days before measurement more than doubled the prediction errors compared to applying the models on data collected after two days. These results demonstrate the importance of keeping full control of the storage time and conditions, when making and applying spectroscopy based models for astaxanthin quantification.

► Optical properties and color of salmon muscle change during storage.
► This effects the prediction of astaxanthin based on spectroscopic measurements.
► Storage time and condition must be controlled when making and applying such models.