Smartphone-Based Colorimetry Analysis for Estimation of Albumin Concentration

The huge surge in smartphone users in the last decade has provided an unprecedented opportunity for App-based diagnosis of infectious and chronic diseases. The internet access has further enhanced the prospects of smartphone-based diagnosis as the data captured can be stored in cloud, analyzed remotely and, the finding may then be placed onto the cloud or communicated through the phone. In this work, commercially available urine strips (Uristix Siemens), chemically impregnated with Bromophenol blue, were used to quantify the concentrations of albumin using a smartphone device i.e. iPhone SE. A Tungsten lamp of color temperature 3500 K was used to illuminate the urine strips. The images of the dipstick after being dipped into different concentrations were captured using a smartphone. The RGB color values were converted into tristimulus values to calibrate the chromaticity curve. Further, the concentration of test samples was calculated using a calibration curve based on a nearest neighbour algorithm. Calibration was done using six different albumin concentrations: 160 mg/L, 320 mg/L, 640 mg/L, 2560 mg/L, 5120 mg/L, and 10240 mg/L. We were able to estimate the albumin concentration in the range of 160 - 10240 mg/L using the proposed algorithm. Three different illuminations i.e. 500 Lux, 400 Lux, and 300 Lux were used to check the robustness of the algorithm. The correlation coefficient for the estimation of albumin concentration was found to be similar to 0.94. This may help to monitor the progression of kidney disease and cardiovascular diseases.