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

• 主要作者: Then, Sing Yick
• 其他作者: Ng Yin Kwee
• 格式: Final Year Project
• 語言: English
• 出版: 2016
• 主題: DRNTU::Engineering
• 在線閱讀: http://hdl.handle.net/10356/67901

The kidney is an organ essential to the urinary system of a body. It is responsible for blood filtration in getting rid of soluble waste material, while maintaining homeostatic functions in a body. Therefore, knowing the health status of the kidney is very much important in the medical field. A form of imaging technique, known as renography is used to diagnose renal obstruction. Due to the absence of standard procedures being applied in clinical setting for such evaluation, this project aims to look at other non-invasive method that diagnoses renal obstruction. From the quantification of renogram, a standard benchmark evaluation for each severity condition can be provided. Hence, the behavior of tracer flow starting from injection to filtration process and later out from the renal pelvis was modeled through compartmental analysis. With the model, mathematical expressions were developed to form critical index, which was compared with doctor’s clinical evaluations. A numerical benchmark for each severity case could then be determined through the comparisons. The index determined was trained on support vector machine (SVM), random forest and adaboost classifier so as to predict the obstruction level of other kidney samples. The predicted results were compared with actual clinical interpretations to identify the predicting accuracy of the classifiers. The performance of the classifiers was further evaluated by Receiver Operating Characteristic (ROC) to obtain a final preferable decision.