Polina Golland

Sandy Wells

PI: Carl-Fredrik Westin, PhD

Multidimensional MRI

PI: Polina Golland, PhD

Statistical Inference for Imaging and Disease

PI: Sandy Wells, PhD

Image Features for Brain Phenotypes

The ability to access huge cohorts of patient medical records and radiology data, the emergence of ever-more detailed imaging modalities, and the availability of unprecedented computer processing power marks the possibility for a new era in neuroimaging, disease understanding, and patient treatment. To unlock the full medical potential made possible by these new technologies, new algorithms and clinically-relevant techniques must be developed by close collaboration between computer scientists, physicians, and medical researchers.

The Neuroimaging Analysis Center presents a national resource center with the goal of finding new ways of extracting disease characteristics from advanced imaging and computation, and to make these methods available to the larger medical community through a proven methodology of world-class research, open-source software, and extensive collaboration. The overarching theme for this P41 is the discovery and analysis of novel imaging phenotypes to characterize disease. We use the term imaging phenotypes to describe patterns or features of disease that can be detected through imaging (predominantly MRI) followed by machine learning, statistical analysis, feature detection, and correlation with other indicators of disease such as structured patient information.

The three presents Technology Research & Development (TR&D) projects address this common question using a variety of complementary approaches and clinical test-beds. TR&D 1 addresses microstructure of tissue, including novel imaging methods to detect tumor microstructure. TR&D 2 investigates rich spatial patterns of disease extracted from clinical imaging with a focus on cerebrovascular and neurodegenerative conditions such as stroke. Finally, TR&D 3 proposes novel image and connectivity-based features that can be correlated with a variety of diseases, with a clinical emphasis on pediatric brain development. Technical innovation will be driven by intense collaboration between the TR&Ds and key collaborators in neurosurgery, neurology, and pediatrics.

The TR&Ds will leverage recent important developments in the fields of image acquisition, machine learning, and data science to identify and exploit novel imaging phenotypes of disease. Building on our long history of developing clinically-relevant methods, each TR&D includes a translational and clinical validation aim to ensure our work is clinically relevant and effective at meeting the driving clinical goals. NAC’s proven software engineering, translation, and dissemination infrastructure, along with its established network of academic, medical, and industrial partners, enhance the center’s value as a national resource.