Automated Imaging Post Processing for MRI DTI Fiber Tracking
Abstract
Purpose
MRI diffusion tensor imaging (DTI) fiber tracking is a valuable tool for visualizing white matter pathways but is typically time consuming and operator dependent. This work presents the development and validation of an automated imaging post-processing tool for MRI DTI fiber tracking.
Methods
We developed an automated post-processing system that connects directly to the scanner PACS, MRI scanners, or DTI DICOM imaging database. The tool processes DTI DICOM images with fully automated distortion correction and image registration. It automatically identifies regions of interest (ROIs) and reconstructs five major white-matter fiber tracts: corticospinal tract (CST), cingulum (CG), inferior longitudinal fasciculus (ILF), superior longitudinal fasciculus (SLF), and anterior thalamic radiation (ATR), using only the input DICOM images of T1 structure and DTI functional data.
Results
To validate the automated processing results, fiber tracking was also performed manually in 5 health subjects for the 5 fibers (CST, CG, ILF, SLF and ATR) . A total of 30 fiber tracts were compared between the manual and automated methods. Of these, 26 fibers demonstrated very good agreement. Manual processing required approximately 2 hours per case to generate all fiber tracts, whereas the automated system completed the same processing in approximately 10 minutes.
Conclusion
The automated DTI post-processing tool is highly efficient and significantly reduces processing time while producing reproducible results that closely match manual fiber tracking. This approach has the potential to improve clinical and research workflows. Further validation with a larger dataset is needed, and future work will include evaluation across disease-specific cases, such as tumor and non-tumor patient populations. Artificial intelligence (AI) technologies will be introduced to identify the human brain fiber tracks more precisely in next version.