We're delighted to report that David Laidlaw and coworkers Eric Ahrens (CMU), John Allman (Caltech), and Mark Bastin (University of Edinburgh) have been awarded a four-year, $2M+ grant from NIH, "DTI+MRI-based tools for analyzing white matter variation". In this multidisciplinary project, a team of investigators will design and apply software tools that can simultaneously segment neural tissues and identify the locations of bundles of neural fibers in the brain. The tools will operate on combined structural and diffusion magnetic resonance (MR) datasets of the nervous system and will produce morphometric measures of each white matter (WM) structure, including its trajectory; cross section, which may vary along the trajectory; and fiber density.
The proposed software tools will globally model imaging datasets. Numerical algorithms will adjust the parameters of a model of neural tissues and WM structures until the model is consistent with all the acquired imaging data and maintains anatomical constraints such as incompressibility and continuity of neural fibers. The tools will differ from current morphometric tools in several ways: they will be more automated; they will incorporate and use all of the complementary information available in the different MR modalities; and they will not have the inaccuracies that are inherent to most current tractography methods.
This research project is innovative in several ways. First, the WM measures will be comparable across subjects without image-level registration because parameters based on WM structures can be compared directly. Second, the investigators will use inverse solution methods to model the multi-valued volume images, globally resolve ambiguities in morphometric measures from local image artifacts such as partial volume effects. Third, the modeling approach will not contract diffusion measurements to tensor values. Thus, hard-to-resolve features such as fiber intersections and projections will be preserved. Finally, the tools will be validated at many levels, including histology, macaque imaging, biological variation in normal volunteers, and clinical feasibility studies in brain tumors, HIV-related neuropathology, and multiple sclerosis. The successful development, validation, and application of these sophisticated software tools may spur further development of medical imaging data modeling. The precise measures of brain structures produced should have a significant impact on biomedical research, will provide a deeper understanding of the brain and how it changes, and could play an important role in surgical planning. More broadly, the tools should apply to research studies of any biological process that involves changes in white matter.
