Urine test strip analysis using image processing for mobile application

Abstract

In attaining vital information for diagnostic purposes in medicine, analysis of the urine sample or urinalysis is a powerful tool. Urine test strip is commonly used in this kind of assessment. In this study, a mobile application has been developed for urine test strip analysis using image processing. Glucose, pH, specific gravity and protein levels are determined and analyzed by the system. The urine test strip is captured using an Android phone, and then the image captured is analyzed using the algorithm employed in OpenCV. Harris detection and RANSAC algorithms are utilized to provide an accurate homography estimation. A portable document format report has been generated to provide the summary of analysis. One-hundred seventy-one (171) urine test strips images are stored in the SQLite database of the mobile phone. And 54 urine samples are tested for the accuracy of the system. The study showed that the overall accuracy is 91.7%.