Quantification of glomerular filtration rate using DCE-MRI in children
Abstract/Contents
- Abstract
- Diagnosis and staging of the chronic kidney disease require the knowledge of the glomerular filtration rate (GFR) of the patients. Although total GFR of a patient can be measured with a blood test, localizing the measurement to a single kidney can only be achieved with the help of radiological imaging techniques. One such technique is dynamic contrast enhanced MRI (DCE-MRI). In this work, we assess the feasibility of using DCE-MRI for localized GFR measurement in pediatric patients. Accurate estimation of GFR through pharmacokinetic models requires concentration-vs-time curves of the contrast agent in blood and renal tissues at a high temporal resolution. However, there is a trade off between the temporal resolution and the spatial resolution in DCE-MRI. In this work, we introduce a post-processing method to estimate the concentration-time curves in the aorta at a high temporal resolution while obtaining images at a high spatial resolution. In order to achieve this goal, a high spatiotemporal resolution DCE-MRI method with view-shared reconstruction was modified to incorporate respiratory-gating, and an aortic input function (AIF) estimation method that use incomplete k-space data from each respiratory period was developed. The methods were validated using realistic digital phantom simulations and demonstrated on clinical subjects. In digital phantom simulations, the HTR-AIF technique had more accurate AIF estimates (RMSE = 0.0932) compared to the existing estimation method (RMSE = 0.2059) that use view-sharing. In the tracer kinetic analysis of the digital phantoms, GFR estimation error using HTR-AIF was less than 10% when the actual GFR was above 27 mL/min. For the same GFR range, the estimation error using view-shared AIF was between 32% and 17%. HTR-AIF method improved both the AIF and GFR estimation accuracies of the respiratory-gated acquisition, and made GFR estimation feasible in pediatric subjects. The concentration-time curves of the aorta and the kidneys needed by the pharmacokinetic models are obtained from the DCE-MRI data using regions of interest (ROIs). The manual segmentation of the kidneys to draw these ROIs could take 2.5 hours per kidney. Semi-automatic segmentation methods (e.g., graph cut) reduce this time to 20 minutes per kidney. In this work, we also introduce a fully automatic renal segmentation method based on iterative graph cut and machine learning techniques that can achieve comparable renal segmentation in less than one minute processing time using the following methods. An image segmentation technique based on iterative graph cuts (GrabCut) was modified to work on time-resolved 3D dynamic contrast enhanced MRI datasets. A support vector machine classifier was trained to further segment the renal tissue into cortex, medulla and the collecting system. The algorithm was tested on 26 subjects and the segmentation results were compared to the manually drawn segmentation maps using F1-score metric. A two-compartment model was used to estimate the GFR of each subject using both the automatically and manually generated segmentation maps. Segmentation maps generated automatically showed high similarity to the manually drawn maps for the whole kidney (F1 = 0.93) and renal cortex (F1 = 0.86). GFR estimations using whole kidney segmentation maps from the automatic method were highly correlated (Spearman rho=0.99) to the GFR values obtained from manual maps. Mean GFR estimation error of the automatic method was 2.23% with an average segmentation time of 45 seconds per patient. The automatic segmentation method performed as well as the manual segmentation for GFR estimation and reduced the segmentation time from several hours to 45 seconds.
Description
| Type of resource | text |
|---|---|
| Form | electronic; electronic resource; remote |
| Extent | 1 online resource. |
| Publication date | 2017 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Associated with | Yoruk, Umit |
|---|---|
| Associated with | Stanford University, Department of Electrical Engineering. |
| Primary advisor | Vasanawala, Shreyas |
| Thesis advisor | Vasanawala, Shreyas |
| Thesis advisor | Hargreaves, Brian Andrew |
| Thesis advisor | Nishimura, Dwight George |
| Thesis advisor | Pauly, John (John M.) |
| Advisor | Hargreaves, Brian Andrew |
| Advisor | Nishimura, Dwight George |
| Advisor | Pauly, John (John M.) |
Subjects
| Genre | Theses |
|---|
Bibliographic information
| Statement of responsibility | Umit Yoruk. |
|---|---|
| Note | Submitted to the Department of Electrical Engineering. |
| Thesis | Thesis (Ph.D.)--Stanford University, 2017. |
| Location | electronic resource |
Access conditions
- Copyright
- © 2017 by Umit Yoruk
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...