Modeling renography data and formulating indices for quantitative means in differentiating kidney obstruction

Abstract

The kidney has a main role in the blood filtration process to get rid of waste materials and maintain homeostatic functions, such as regulation of electrolytes, maintenance of acid-base balance and regulation of blood pressure. Renography is a kidney imaging technique used to detect renal health status. However for the purpose of diagnosis renal obstruction, there is still no precise technique and standard protocol accepted and applied in the clinical setting. This project was carried out to search for a non-invasive method in the assessment of renal obstruction and to come out with a benchmark for clinical evaluation of the severity of obstructed kidney. In order to achieve this objective, the model that represented the behaviour of tracer from the input into kidney through filtration process to the flow out from the renal pelvis was developed using two compartmental modelling. Then, the model was compared to clinical data from renography and it had been verified in this project that the mathematical model was accurate in predicting the relative severity of obstructed kidney. Lastly, using support vector machine (SVM) classifier as a quantitative means for differentiating kidney obstructions was proposed based on the simulation results of the samples that had been compared with clinical interpretation of renograms by a certified nuclear medicine doctor.