Using a Smartphone for Point-of-Care Urinalysis of Chronic Kidney Disease

Author: Shaymaa Akraa

Akraa, Shaymaa, 2018 Using a Smartphone for Point-of-Care Urinalysis of Chronic Kidney Disease, Flinders University, College of Science and Engineering

Terms of Use: This electronic version is (or will be) made publicly available by Flinders University in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. You may use this material for uses permitted under the Copyright Act 1968. If you are the owner of any included third party copyright material and/or you believe that any material has been made available without permission of the copyright owner please contact with the details.


The main objective of this research is using a smartphone to provide early detection and periodic monitoring of chronic kidney disease (CKD) by analysing the human serum albumin (HSA) in urine using aggregation-induced emission (AIE) bioprobes. The intensity of AIE probe fluorescence increases as the HSA concentration increases. This knowledge is exploited by preparing precise graded concentrations of HSA that cover the five levels of CKD. AIE bioprobes are used to enhance the fluorescent intensity of the HSA solution. The solution is imaged within an imaging housing that ensures the same imaging conditions for all samples. The captured images are transformed from the red-green-blue (RGB) colour space into the hue-saturation-luminance (HSL) colour space to extract the intensity of the images. These intensity values are used to derive a linear regression between image intensity mean values and the corresponding HSA concentration. An important stage in the imaging process is the enhancement of the relation between image intensity and HSA concentration in order to avoid distortion. The colours of images that have been captured by one smartphone camera may not be equivalent to those that have been captured by other smartphones’ cameras. The intensity values of images cannot be identical across various smartphone cameras due to differences in sensor response; therefore, a calibration process is applied to get a united intensity relation. The images are also affected by gamma correction that is applied by the smartphone's camera software after they are captured. This results in the intensity value relation to be non-linear. As the gamma values used during gamma correction is unknown or unpublished by the smartphone manufacturers, a blind inverse gamma correction technique is applied in order to estimate the unknown gamma values so that gamma corrected images are converted back to the raw images from which the original intensity values are extracted. Principle component analysis (PCA) is applied to calculate the linear regression between image intensity and HSA concentration, while K-fold supervised machine learning, where the best model is picked based on mean square error (MSE), is applied to predict the linear regression using the Least Squares Method. Both PCA and K-fold machine learning create an equivalent linear regression with trivial error. The predicted linear regression is used to determine the HSA concentration corresponding to the intensity value of the captured image.

Keywords: self monitering by smartphone

Subject: Computer Science thesis

Thesis type: Doctor of Philosophy
Completed: 2018
School: College of Science and Engineering
Supervisor: Dr. Haifeng Shen