Smartphone-Based Urine Strip Analysis

Point-of-care testing (POCT) has transformed the healthcare landscape by delivering quick and cheap diagnostic services closer to the patient. Urine test strips are one of the most commonly used POCT tools, however their manual interpretation can be challenging, particularly for the elderlies and people with eye disorders. In this study, we propose a smartphone application designed to automatically perform semi-quantitative colorimetric analysis on urine strips by using just one image of the strip, placed on a specially designed reference card. Virtually no shooting restrictions apply while the system is able to adapt to different smartphones and varying ambient light conditions. For the detection of the card, we match ORB keypoints between the captured image and the reference card in memory, and use RANSAC to calculate the projective transform. Then, the strip's rectangular pads are precisely detected and their chromaticity CIELUV components are compared to the manufacturer's reference after being corrected using SVM regression. We tested the application on the analysis of pH values, using three different smartphones on various lighting conditions and obtained promising results that prove its concept. Future work includes the extension of the system to perform fully quantitative analysis to additional analytes and colorimetric strips.