Description:

Abstract

Neuroanatomists describe white-matter fiber bundles based on the anatomical regions that the bundles go through or next to. In this talk I will describe methods for mapping white-matter pathways from diffusion MRI data that take a similar approach; that is, they define pathways based on their anatomical neighbors, rather than coordinates in a template space. The benefit of this approach is robustness to individual anatomical variability, allowing us to analyze data in the presence of disease and across the human lifespan.

First, I will discuss a method for automated reconstruction of known white-matter pathways that combines global probabilistic tractography with prior information on the neighboring anatomical structures of the pathways. I will show our recent extensions of this method to handle longitudinal data, and to map pathways in the infant brain. I will present results suggesting that this approach is robust to differences between the study subjects and subjects that were used to train the algorithm. I will also discuss our efforts to extend the idea of defining bundles based on their anatomical neighborhood to an exploratory approach that does not rely on prior information. Finally, I will suggest a path towards augmenting our prior knowledge on brain pathways based on multimodal ex vivo data at microscopic resolutions, with the ultimate goal of improving the reconstruction of these pathways from in vivo data at routinely available resolutions.

Speaker

Dr. Anastasia Yendiki, PhD is currently Assistant Professor of Radiology, Harvard Medical School, and Assistant Physicist, Massachusetts General Hospital.
Dr. Yendiki's background is in statistical signal and image processing. She received a PhD in Electrical Engineering: Systems from the University of Michigan at Ann Arbor in 2005, where she worked on inverse problems in tomographic reconstruction for nuclear imaging under the supervision of Jeff Fessler. As a postdoctoral research fellow at the Athinoula A. Martinos Center for Biomedigal Imaging, Harvard Medical School and Massachusetts General Hospital, she trained in functional and diffusion-weighted MRI. She is the lead developer of TRACULA, the diffusion-weighted MRI analysis stream in FreeSurfer, under the supervision of Bruce Fischl. She is now faculty at the MGH/HMS Martinos Center and a member of the Laboratory for Computational Neuroimaging (LCN), continuing to develop publicly available, open-source algorithms for studying white-matter anatomy in health and disease.