Determination of serum protein content using cell phone image analysis

Abstract

This paper presents a simple, fast and inexpensive way to measure serum protein content (albumin and total proteins) by integration of color images acquired with a cell phone camera and multiple linear regression (MLR). MLR models were built using the RGB-HSV image systems from the color changes induced by bromocresol green and Biuret methods on solutions containing albumin (0.24 to 5.03 g dL⁻¹) and total proteins (0.25 to 8.20 g dL⁻¹), respectively, acquired from a cell phone camera. These methods were adapted to a microscale analysis by using an ELISA 96-microwell plate of 250 μL as a reaction container, and subsequently were statistically validated. The RMSEP values obtained for albumin and total protein models were respectively equal to 0.28 and 0.12 g dL⁻¹, and their precision was respectively equal to 6.16 and 4.85%. No statistical difference was observed at a confidence level of 95% between the protein concentration calculated by the proposed method and those found by the reference method, proving the reliability of the imaging method as an alternative approach for these assays.