Record:   Prev Next
Author Song, Ting
Title Optimization of MR protocols for spatial-temporal analysis of four-dimensional dynamic renal images
book jacket
Descript 199 p
Note Source: Dissertation Abstracts International, Volume: 68-11, Section: B, page: 7480
Adviser: Andrew F. Laine
Thesis (Ph.D.)--Columbia University, 2008
Imaging plays an essential role in the diagnosis and treatment of renovascular disease (RVD). MR imaging with gadopentetate dimeglumine (Gd-DTPA) can provide a non-invasive method for the measurement of single kidney function. Four-dimensional (4D) MR renography has the potential to supplement anatomical renal MRI examinations because it has sufficient spatial, temporal and contrast resolution when used with contrast-enhanced dynamic protocols. Quantitative analysis of 4D MR renography provides a more objective and accurate approach to diagnosis of RVD compared to visual analysis. This dissertation develops a novel quantitative framework including acquisition optimization, automated 4D image registration, and automated 4D image segmentation for accessing renal function
In order to optimize MR protocols for fast time-resolved acquisitions for 4D dynamic renal images, a model was developed for selecting imaging parameters based on minimization of errors in signal intensity versus time and physiologic parameters derived from tracer kinetic analysis. Optimization was performed for time-resolved angiography with a stochastic trajectory (TWIST) algorithm applied to MR renography. High-quality dynamic human images were acquired with optimal TWIST parameters and 2.5 seconds temporal resolution. This method can be generalized to other dynamic contrast enhanced MRI applications, including cancer imaging, and under-sampling schemes
4D renal images are affected by respiratory motion that limits analysis and interpretation. Each examination yielded 41 3D images of the abdomen, making manual registration prohibitively labor-intensive. Even with repeated end-expiratory breath holds, in-plane translation, out-of-plane motion and rotation are observed across images within a series. In this thesis, a fully automated technique that corrects for translation and rotation motion in 4D MR renography is presented. A method using Wavelet Representations and the Fourier Transform (WRFT) was first evaluated on a set of simulated MR renography images with defined degrees of kidney motion
After registration, segmentation of each intrarenal compartment is required, a task that is prohibitively labor intensive for clinical practice. Existing methods do not fully utilize the wealth of information in 4D data sets. In this context, a novel 4D segmentation framework based on temporal dynamic 4D level sets was proposed to fully combine temporal dynamics and spatial information, with less than one minute needed to automatically segment a one gigabyte 4D data set. Compared with manually segmented results, both volume and signal intensity errors were statistically comparable to the range of inter-observer variance between two experienced radiologists
The image analysis methods proposed in this thesis developed for 4D MR renographic images, can easily be extended to other applications involving 4D image series. To illustrate the flexibility of the proposed methods, an application to 4D lung tumor perfusion MR images was also presented. (Abstract shortened by UMI.)
School code: 0054
DDC
Host Item Dissertation Abstracts International 68-11B
Subject Engineering, Biomedical
0541
Alt Author Columbia University
Record:   Prev Next