Rapid quantification of oxidation indicators in fish oil by digital image colourimetry

Abstract

A digital image colourimetry (DIC) method was developed to rapidly determine the acid value (AV), peroxide value (PV) and iodine value (IV) of fish oils. The copper-soap, iodometric, and Wijs reactions were adapted for smartphone imaging by performing the reactions in light-isolated boxes and setting the acquisition settings to ensure consistent colour capture. Colour features from RGB/Lab/YUV spaces were extracted and screened using the Pearson correlation coefficient to retain only the most informative channels for AV, PV and IV. Six regression algorithms were compared, and the random forest model showed the highest predictive accuracy. The method exhibited acceptable repeatability and low detection limits, and it agreed closely with titration results, with a correlation coefficient above 0.99. Relative to titration, the DIC approach reduced reagent consumption by at least 94%, shortened the analysis time per sample by at least 84%, and reduced the required sample mass to 0.02–1.0 g. It also avoided the use of ether and chloroform employed in some conventional titration procedures, although cyclohexane, isooctane, and glacial acetic acid were still required. DIC thus provides a rapid, low-cost, and more resource-efficient alternative for routine screening and quality control of fish oil oxidation.