Grant: Discovery Grants Program - Individual

The severity of the crisis of reproducibility in science has led to a growing appreciation of the importance of building open data standards and related resources that enhance the transparency of methods used for scientific discovery. Although open data standards exist in neuroimaging, none of the existing implementations have been proven to enable localization at the millimetric or sub-millimetric scale, which is the degree of spatial specificity necessary for the detection of subtle variations in brain structure as well as applications like deep brain stimulation (DBS). Stereotactic neurosurgery is a subspecialty within the practice of neurosurgery concerned with accurate targeting of brain structures for diagnostic and therapeutic purposes including DBS. Two key components allow for accurate stereotactic mapping, (1) a stereotactic space i.e. anatomical data prepared in Cartesian (x, y, z) coordinates and (2) a mapping function that establishes a correspondence between the stereotactic space and the brain. These principles have been employed in neurosurgery to facilitate refined targeting of focal brain structures. Unfortunately, this level of accuracy is not typically attained in neuroimaging studies thus limiting sensitivity to focal discoveries. To complicate matters further, a multitude of extant stereotactic spaces are used by the community and conversions between these spaces contribute to further inaccuracies. Inspired by these stereotactic principles, we developed the Anatomical Fiducials (AFIDs) framework to help address this gap in natural science and engineering applications, defining salient anatomical features visible using conventional structural magnetic resonance imaging (MRI) scans that can be reliably placed at the millimetric scale. Our initial work involved the creation and validation of this framework in cognitively normal adult human brains. This entire workflow has been created with full transparency in mind so that others may freely use, adopt, and modify including the adoption of the Findable, Accessible, Interoperable, and Reusable (FAIR) principles. To this end, associated datasets and analysis workflows have been released open source on GitHub (https://github.com/afids). This research program involves the development and maintenance of open data standards for focal description of neuroimaging discoveries, development of tools for automatic detection of visible (and invisible) neuroanatomical structures, as well as using this paradigm to identify subtle morphological changes associated with normal aging. This program of research is inherently multidisciplinary involving medical image analysis and processing, machine learning, neuroanatomy, and methods used in stereotactic neurosurgery. The developments and findings from this research program will facilitate aggregation of data from neuroimaging studies that is both transparent and accurate paving the way for increasingly focal discoveries in health and disease.