Computation of statistics for populations of diffusion tensor images.
详细信息
- 作者：Goodlett ; Casey Brett.
- 学历：Doctor
- 年：2009
- 毕业院校：The University of Utah
- ISBN：9781109246209
- CBH：3364929
- Country：USA
- 语种：English
- FileSize：1897544
- Pages：159
文摘
The magnetic resonance imaging MRI) technique known as diffusion tensor imaging DTI) provides a unique method for investigating the architecture of neural white matter in human subjects in-vivo. DTI has a promising ability to better understand the biological root of disease in white matter between healthy subjects and those presenting white matter pathology or altered cognitive function. Because of the complexity of the acquired data, new image analysis techniques are essential to enable clinical research. This thesis presents a framework for evaluating tract specific group differences in populations of diffusion tensor images. The ability of DTI to quantify diffusion parameters in living tissue requires a careful understanding of the imaging protocol and techniques for estimating diffusion parameters from measurements. An investigation of the effect of imaging noise on both tensor estimation and gradient sequence design is presented using both simulation and validation experiments. This evaluation provides recommendations for future studies as well as an understanding of potential confounds in retrospective analysis. Techniques for atlas building and tract-based analysis are developed to provide a reference coordinate frame for statistical analysis. Atlas building enables the study of a population of images in a common coordinate system. Novel validation measures for comparing streamline tractography results are used to evaluate the results of atlas tract identification. Tract-based analysis of diffusion measures within atlas space enables intuitive statistical methods for testing the differences of specific tracts. The statistical framework allows for joint analysis of multiple diffusion statistics and accounts for along tract correlation. These methods provides a generic framework for neuroimaging studies and has been demonstrated on clinical studies of normal development and schizophrenia to illustrate potential new findings as well as confirmation of previous studies.