Our research focuses on diffusion magnetic resonance imaging, the only method that can map the connections of the living human brain. We have three main research areas: novel computational methods, open-source software, and neurosurgical planning.
Novel Computational Methods
Dr. O'Donnell's group investigates methods for machine learning and statistical analysis based on data from diffusion MRI fiber tractography. Our method for fiber clustering is a machine learning technology that enables discovery of thousands of unique white matter brain connections that are found very robustly in large groups of subjects. We refer to this as data-driven white matter parcellation. This technique enables neuroscience and neuroanatomy research, as well as research in neurosurgical planning. Recent studies include investigations of autism using machine learning classification techniques, novel statistical analyses that leverage the geometry of the fiber tracts, and ongoing work to anatomically annotate and publicly release curated white matter fiber cluster atlases.
Our NIH-funded open-source software, SlicerDMRI, is downloaded 200 times per month and used in multiple brain research studies for diffusion MRI visualization and analysis. Our software for diffusion MRI fiber tractography clustering, whitematteranalysis, is available as open source.
Dr. O'Donnell leads Aim 1 of the Neurosurgery Core of the BWH P41 Ferenc Jolesz National Center for Image-Guided Therapy. Recent research includes the automated detection of crucial fiber tracts in patients with brain tumors, as well as investigations into tractography methods for neurosurgery. This has the potential to improve preservation of crucial white matter fiber tracts during surgery.