The following projects are geared towards understanding existing research methods and creating more accurate research tools in fMRI.
Empirical Power Analysis Tool for fMRI
R00 phase funded by NIMH R00MH130894
PI: Stephanie Noble
K99 phase funded by K99MH130894 PI: Stephanie Noble
Recent work has exposed an endemic lack of statistical power (i.e., ability to detect effects of interest) and a need for power analysis tools that meet the demands of the typical user. We propose to create a web-based power calculator tailored to typical fMRI studies that only requires the user to specify information readily available to them. By enabling researchers to more easily and accurately plan studies for desired levels of power, this power calculator will promote more robust and reproducible findings in the field.
The Constrained Network-Based Statistic: A New Level of Inference for Neuroimaging
Funded by NIMH K00MH122372
PI: Stephanie Noble
Current multiple comparison correction procedures for functional connectivity inference do not account for dependence between non-contiguous connections or spatial bias arising from differences in measurements across the brain. This project seeks to address these limitations with a new a priori network-based multiple comparison correction approach, establish its empirical utility in the context of other correction approaches, and maximize its impact through modern open science practices.
Improving Reliability and Validity of fMRI Statistical Methods
Funded by NINDS F99NS108557
PI: Stephanie Noble
Recent work has raised questions about the reliability and validity of major fMRI statistical methods. The first part of this project addressed open questions regarding the test-retest reliability of functional connectivity. The second part of this project evaluated sensitivity of common cluster-based inference methods used in activation mapping.
BioImage Suite Web
I am part of the BioImage Suite Web (BISWeb) team led by Dustin Scheinost and Xenophon Papademetris (R24 MH114805, PIs: Papademetris X. and Scheinost D). BISWeb is a point-and-click app that runs in any modern web browser without any installation necessary, yet with performance rivaling that of a locally installed software. Due to the general nature of its architecture (C++, WebAssembly, JavaScript), BISWeb works on most modern web-browsers, major operating systems, and devices. Currently, BISWeb supports a range of functionality in human and animal neuroscience, from defacing to connectome visualization.
