Evaluation of Color Models for Quantitative Determination of Food Dyes using Smartphone-Based Digital Image Analysis

Giovania Evangeline Halim (Department of Chemical and Food Processing, Calvin Institute of Technology, Jakarta, Indonesia Calvin Tower RMCI Jl. Industri Blok B14, Jakarta)
Martin Tjahjono (Department of Chemical and Food Processing, Calvin Institute of Technology, Jakarta, Indonesia Calvin Tower RMCI Jl. Industri Blok B14, Jakarta)

JKPK (Jurnal Kimia dan Pendidikan Kimia)
Vol 8, No 1 (2023): JKPK (Jurnal Kimia dan Pendidikan Kimia)

In recent years, smartphones for digital image analysis (DIA) have emerged as an affordable, user-friendly, and accessible chemical and food analysis tool, particularly in colorimetry. This study aimed to compare the performance of various color models and demonstrate their usefulness in quantifying food dyes in commercial products using DIA. Images of food dye solutions at 500 Lux were captured using an OPPO F11 smartphone, and the RGB values are mathematically transformed into several color models. The results show that the normalized blue channel was the most robust color model for analyzing different food dyes using DIA. The corresponding limit of detection (LOD) and limit of quantification (LOQ) for nine food dyes studied are following: carmoisine, 3.7 and 11.3 mg/L; sunset yellow, 1.0 and 3.1 mg/L; allura red, 2.0 and 6.0 mg/L; ponceau 4R, 1.3 and 4.0 mg/L; tartrazine, 5.0 and 15.2 mg/L; fast green, 2.0 and 6.1 mg/L; brilliant blue, 1.9 and 5.7 mg/L; quinoline yellow WS, 3.3 and 9.9 mg/L and indigo carmine, 1.2 and 3.8 mg/L. These LOD and LOQ values were comparable to those obtained from UV-Vis spectroscopy measurements: carmoisine, 2.4 and  7.2 mg/L; sunset yellow: 0.9 and 2.6 mg/L; allura red, 1.4 and 4.2 mg/L; ponceau 4R, 1.9 and 5.7 mg/L; tartrazine, 0.9 and 2.7 mg/L; fast green, 1.5 and 4.4 mg/L; brilliant blue, 3.6 and 10.9 mg/L; quinoline yellow WS, 0.3 and 0.9 mg/L and indigo carmine, 4.3 and 13.0 mg/L. The DIA method was successfully applied to determine the concentrations of food dyes in three commercial samples (Samples S1-S3) containing carmoisine, tartrazine, and brilliant blue, respectively. The measured concentrations are 52.7±2.6 mg/L (S1), 105.9±5.4 mg/L (S2) and 7.9±0.5 mg/L (S3), which are in good agreement with UV-Vis spectroscopy results employing standard addition method 58.2±3.0 mg/L (S1), 106.2±1.3 mg/L (S2), 8.3±0.5mg/L (S3). Overall, this color model study demonstrates the utility of DIA method as a reliable and affordable food dye analysis tool that can potentially be used for public health and safety monitoring.