FPGA-Based Urinalysis Using Principal Component Analysis

Abstract

Urinalysis is considered to be a common test performed in laboratory in order to diagnose Urinary Tract Infection (UTI). It undergoes three stages, which include macroscopic, dipstick, and microscopic analysis. This paper describes a way of performing urinalysis for UTI detection using the Principal Component Analysis (PCA) implemented using a Field Programmable Gate Array (FPGA). Input to the system is from five ion-selective sensors that measure five different components specifically sodium, nitrite, nitrate, potassium, and pH level of a urine sample. Tests show that the system obtained an accuracy of 94.13% for standard urinalysis showing the accuracy of sensors and measurements used. To be able to detect the presence of UTI in urines, an outlier detection method Principal Component Analysis (PCA), was used. PCA is a tool used in reducing multidimensional data to lesser dimensions while keeping all the information. An accuracy of 83.33% in detecting UTI infection was achieved. The accuracy of FPGA implementation of PCA was compared with MATLAB calculation results and an accuracy of 99.917% was achieved.