The objective is to explore a novel data processing strategy that would represent a technological breakthrough toward improving the capabilities of clinically acquired Diffusion Tensor Imaging (DTI) data using a Magnetic Resonance Imaging (MRI) scanner. If successful, the proposed approach would lead to dramatic improvements in the visualization and quantification of brain connections, and any abnormality therein, using DTI data acquired by commonly used protocols in clinical scanners. As cerebral connections depend critically on the integrity of white matter, it is clinically important that these connections be established accurately. The prevalent DTI based streamline tractography approaches rely on the assumption that there is only one fiber per voxel. However, when multiple fibers (or tracts) cross in a voxel, the primary tract orientation obtained by DTI is likely to point in an erroneous direction -- biased toward the highest density fibers -- causing many tracts to propagate in the wrong direction leading not only to incorrect or incomplete identification of brain connections and pathways but also erroneous DTI metrics used in the diagnosis. Most existing methods to solve the so-called multiple-fiber or fiber-crossing problem generally require a relatively long time to acquire diffusion-weighted data, which, in combination with the other required anatomical MRI scans, are generally considered impractical for clinical use due to patient movement and total time in the scanner. Several of the existing methods to incorporate multiple fibers per voxel also require specialized data acquisitions protocols not readily available at many clinical sites. Here we propose a novel Independent Component Analysis (ICA) based approach that would provide a simple, rapid and viable alternative to single-fiber tractography by enabling multiple-fibers per voxel to be identified and incorporated in the tractography. The proposed approach would be applicable to clinically acquired diffusion weighted data without requiring any specialized pulse sequences. Moreover, the approach could be readily implemented to process already existing routinely acquired clinical DTI data in many labs for a variety of clinical studies. After simulation and validation studies with a phantom to develop the methodology, we will apply ICA to existing human clinical DTI data acquired from a cohort of Alzheimer Disease subjects using a 3T MRI and another from a cohort of Traumatic Brain Injury subjects at 1.5T to evaluate the performance of ICA in these two clinical situations.

Public Health Relevance

Novel ICA Based Multi-Fiber Streamline Tractography Approach Narrative The objective is to develop a novel method to improve diffusion tensor imaging (DTI) tractography for clinical use. DTI is a Magnetic resonance imaging (MRI) based method widely used to study brain connections. Brain connections provide invaluable information on normal and abnormal brain function such as functional changes that may occur in Alzheimer Disease or traumatic brain injury. Most current processing strategies for clinically acquired DTI data are limited to determining only a single white-matter fiber or one connection-direction per location (or voxel) in the brain, whereas the proposed new method will incorporate multiple connections or multiple white-matter fibers in a voxel. This will allow more realistic tractography to be conducted with existing clinical human data, or data acquired in ongoing or planned clinical studies, to visualize and quantify brain connectivity more accurately.

National Institute of Health (NIH)
National Institute of Biomedical Imaging and Bioengineering (NIBIB)
Exploratory/Developmental Grants (R21)
Project #
Application #
Study Section
Special Emphasis Panel (ZRG1-NT-L (09))
Program Officer
Liu, Guoying
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
University of Southern California
Schools of Medicine
Los Angeles
United States
Zip Code
Huhdanpaa, Hannu; Hwang, Darryl H; Gasparian, Gregory G et al. (2014) Image coregistration: quantitative processing framework for the assessment of brain lesions. J Digit Imaging 27:369-79