Image Processing Considerations for High Resolution Diffusion Tensor Imaging of the Human Brainstem




Hulsey, Keith McLeod

Journal Title

Journal ISSN

Volume Title


Content Notes


Diffusion weighted MRI is used to measure the diffusivity of water in the human brain noninvasively. Diffusion tensor imaging (DTI) fits the diffusivity measurements from many directions to a tensor model of the diffusivity of water in brain tissue. DTI is particularly useful for interrogating the health and organization of white matter in the brain. The human brainstem has many white matter tracts that connect small nuclei in the brainstem to other regions of the brain. High resolution DTI of the brainstem may be helpful in understanding diseases that implicate brainstem nuclei. There are technical challenges for DTI which must be addressed to provide the most sensitive and meaningful measurements. Some of these challenges are: accurate registration between diffusion weighted images, accurate fitting of the data to the tensor model, measuring significance of group differences using DTI results, and increasing image resolution. This study has focused on finding solutions for accurate fitting of data to the tensor model in the presence of signal void artifacts and on increasing image resolution beyond the point at which signal aliasing occurs. To meet the aims of this study I have; 1) developed an innovative approach to detect and remove signal void artifacts caused by subject motion, brain motion induced by cardiac pulsation and scanner vibration, 2) developed an innovative approach to mask aliased signal in DTI scans which have a field of view smaller than the subject's head, and 3) shown that removing signal void artifacts from the DTI scans acquired for the Gulf War Illness study produces significant changes in FA for most subjects in the study. Removing signal void artifacts from the Gulf War Illness study data did not, however, alter the conclusions of group comparisons for the samples of Gulf War veterans studied. Two conclusions of this study are that signal void artifacts should be removed from DTI data before conducting analysis and that an image with a field of view larger than the subject's head can be used to estimate the location of aliased signal in DTI scans acquired with fields of view smaller than the subject's head.

General Notes

Table of Contents


Related URI