I am a fourth year PhD student in Biostatistics in the Department of Biostatistics, Epidemiology, and Informatics at the University of Pennsylvania under the mentorship of Taki Shinohara. I obtained my undergraduate degree in psychology from Pennsylvania State University some years ago and have been working in neuroimaging since I graduated.
My latest statistical interests are in multiple testing procedures for high-dimensional dependent data. My work is motivated by problems in neuroimaging and psychology, and I am particularly interested in applications in those fields.
Cortical Coupling is a measure we created to describe localized relationships between characteristics of the cortex. The work was published in Neuroimage in 2016 with code to estimate coupling from Freesurfer volumes. My advisor and colleagues have extended the work to improve segmentation of MS lesions, and make coupling estimates symmetric between modalities and include more than two modalities.
The coupling paper was an extension of previous work where our goal was to investigate the relationship between cortical characteristics as the brain matures through adolescents. In this work, we developed code to perform spatial permutation tests to assess the spatial coherence between two measures on the cortical surface. Spatial permutation tests are appropriate for data that can be projected onto a sphere.
We have been working on several approaches to hypothesis testing that aim to control the FWER when performing tests at each voxel or a set of brain regions. The approaches we implement account for the strong dependence structure between the test statistics in neuroimaging. Several recent studies have shown that mutliple testing procedures for neuroimaging that rely on Gaussian random field theory have inflated FWERs. In my latest paper we proposed a fast testing procedure for region- and voxel-wise analyses that controls the FWER at the nominal level. We also compared the finite sample properties of parametric and permutation procedures. My future work aims to develop fast and valid cluster-based inference.
For a complete listing of my publications please check out my Google Scholar Page.