Strip Test Analysis Using Image Processing for Diagnosing Diabetes and Kidney Stone Based on Smartphone

摘要

Urine is a useless substance that produced by the body metabolism that needs to be excreted. Many research about urinalysis shows that the excreted urine can be an indicator of someone's health condition by the composition of urine itself. One of the methods used to know about urine composition is by using strip test. Strip test works similarly with litmus paper, when the surface of the paper is making a contact with the substance, the color of the reagent will change. Strip test can show the user about the substance that should not be excreted alongside the urine, for example glucose, blood, nitrite, ketones, bilirubin, and urobilinogen. The problem of using a strip test is that without any strip test analyzer device (semi-reflective photometer), some people will have difficulties interpreting the result with the reference table [1]. This system is created to help to interpret the strip test result using image processing by capturing the strip picture using a smartphone. The system can determine the discoloration of the strip with more details and could give a result whether the patient is possibly diabetes or not and detect the possibilities of kidney stone. From the results of the experiments conducted, the system is able to classify the result with 93.4 % accuracy for blood classification (to identify the kidney stone) and 90% accuracy for glucose classification (to identify diabetes). It can be concluded that the system can be used to carry out strip test without any help of semi-reflective photometer or medical expert.