BrainEffeX: A Shiny app to explore typical effect sizes in functional neuroimaging research

Poster Session C - Sunday, April 14, 2024, 5:00 – 7:00 pm EDT, Sheraton Hall ABC

Hallee Shearer¹ (h.shearer@northeastern.edu), Matt Rosenblatt², Jean Ye², Rongtao Jiang², Link Tejavibulya², Qinghao Liang², Javid Dadashkarimi², Margaret Westwater², Iris Cheng², Max Rolison², Hannah Peterson², Brendan Adkinson², Saloni Mehta², Chris Camp², Joshua Curtiss¹, Dustin Scheinost², Stephanie Noble^1,2; ¹Northeastern University, ²Yale University

Typical functional MRI (fMRI) studies are often statistically underpowered, leading to the inability to replicate many findings. This can arise when power analyses are conducted improperly or skipped altogether. One reason power analyses are so difficult is that it is currently not straightforward to estimate expected effect sizes from existing literature. To address this, we created a Shiny app to facilitate exploration of effect maps derived from “typical” fMRI studies. Using six large (n>500) publicly available fMRI datasets (ABCD, HCP, HBN, SLIM, PNC, UKB), we created a database of 38 effect size maps for frequently used study designs encompassing functional connectivity (FC) and task-based activation, then conducting within-subject, between-group, and correlation analyses for FC and activation. We then created a user-friendly Shiny app to explore these maps. Users can select a map type (FC or activation) and test type (1-sample, 2-sample, correlation), and optional inputs (e.g., specific dataset, task, or behavior of interest). The app then displays Cohen’s d values alongside 95% simultaneous confidence intervals across edges or voxels for studies that fit the selected criteria. Effect maps are also visualized as connectivity matrices (FC studies) and mapped onto the brain (activation studies). The database and explorer are intended to be the starting place for an evolving resource to define some guidelines for effect sizes we may expect in the community. Furthermore, these resources will facilitate the ongoing development of an empirical power calculator for fMRI studies needed to facilitate study planning for more robust and reproducible neuroimaging research.

Topic Area: METHODS: Neuroimaging