Modelling renography data and formulating indices for quantitative means in differentiating kidney obstructions

• Main Author: Gong, Ying Ying
• Other Authors: Ng Yin Kwee
• Format: Final Year Project
• Language: English
• Published: 2015
• Subjects: DRNTU::Engineering::Mechanical engineering
• Online Access: http://hdl.handle.net/10356/63467
• Institution: Nanyang Technological University

The kidney acts as a system of purifying blood and removing metabolism waste products is a very important organ of human body. Kidney with obstruction will be failing after a few weeks. Modern medicine applies renography technique to detect kidney issues as well as the renal obstruction diagnosis. In this technique, a tracer is introduced into the blood circulation. To capture the image of kidney, the amount of tracer is measured by radioactive means. However, it is a potentially invasive method and doesn’t have standardized protocols and diagnostic criteria. This project aims are to model the tracer behaviour from input to the washout from the renal pelvis and compares with the clinical data detected by renography. To achieve this, the mathematical model was carried out in this project. Moreover, obtain a benchmark for clinical evaluation of the severity in obstructed kidney. Support Vector Machine (SVM) classifier was used to predict and formulate indices for quantitative means in differentiating kidney obstructions. Random Forest classifier was also proposed to compare the simulation results of the samples with SVM classifier. It has been verified in this project that Random Forest allowed more accurate analysis to the clinical interpretation of renograms from a certified nuclear medicine doctor in distinguishing the level of the severity for obstructed kidney.