Diffusion tensor imaging application

Litao Shen, Purdue University


Central nervous system (CNS) related conditions and diseases like mild traumatic brain injury (mTBI) and multiple sclerosis (MS) affect people’s life quality, yet there is no single test for the diagnosis of these diseases or conditions. Patients may need to wait for years until they are diagnosed correctly to get the correct treatment, which is often too late. Thus, there is a strong need to develop some techniques to aid the diagnosis of CNS-related conditions and diseases. The conventional MRI can reveal the structure of the brain but cannot detect the difference between the healthy tissue and the anomalies. Diffusion tensor imaging (DTI) has been used for detecting white matter integrity and demyelination for the past decade in experiments and has been proven to have the ability to depict the problem effectively. In the past decade, many techniques were found based on DTI data, and these techniques improved pre-processing, processing, and post-processing. Though there are many software and APIs that can provide functions for DTI file input/output (IO), visualization and other DTI related topics, there is no general software or API that is dedicated to covering the whole processing procedure of DTI that at the same time can be extended easily by the user. This thesis is dedicated to developing a software that can be used to aid in the diagnosis of CNS-related conditions and diseases while at the same time trying to cover as many topics as possible. Another purpose is to make the software highly extensible. This thesis work first introduces the background of CNS-related disease and uses MS as an example to introduce the process of demyelination and the white matter integrity problem, which are involved in these CNS-related diseases and conditions. Then the diffusion process and the technique that can detect the diffusion signal (DTI) is presented. After this, concepts and meaning of the secondary metrics are discussed. Then, current existing software and APIs and their advantages and disadvantages are outlined. After these points, the techniques that are discussed in this thesis as well as their advantages are outlined. This part is followed by the charts and code samples which can illustrate the process and structure of this software. Then different modules and their results are explained. In this software, the results are represented by images and 3D models. There are color images, pseudo color images with different schemes and gray scale images. Images are mainly included to represent the FA and MD data. In this software, streamlines are generated from the eigenvalue and eigenvector. Then a bundled result for the streamline is also realized in this software. The streamline and bundled results are 3D models. For 3D models, there are mainly two ways to display the real 3D model. One is the naked eye 3D which doesn’t require the user to wear glasses but has less stereoscopic characteristics. As the stereoscopic monitors and glasses are more and more popular and easily accessible, this software also provides stereoscopic views for 3D models, and the user can choose red & blue, interlaced techniques with proper glasses. This thesis work ends with the discussion of the results and limitations of DTI. Finally, there is a discussion about the future work that can improve the performance of this software and topics that need to be covered.




Chen, Purdue University.

Subject Area

Biomedical engineering|Medical imaging

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server