Schedule of Events | Search Abstracts | Invited Symposia | Symposia | Poster Sessions | Data Blitz

Postdoctorial Fellowship Award Winner 

PRISME: A MATLAB Toolbox for Multi-Method Statistical Power Analysis in Neuroimaging

Poster Session B - Sunday, March 8, 2026, 8:00 – 10:00 am PDT, Fairview/Kitsilano Ballroom
Also presenting in Data Blitz Session 4 - Saturday, March 7, 2026, 10:30 am – 12:00 pm PST, Salon F.

Fabricio Cravo¹, Alex Fischbach¹, Hallee Shearer¹, Stephanie Noble^1,2,3,4; ¹Department of Psychology, Northeastern University, Boston, MA, ²Department of Bioengineering, Northeastern University, Boston, MA, ³Center for Cognitive \& Brain Health, Northeastern University, Boston, MA, ⁴Department of Radiology \& Biomedical Imaging, Yale University, New Haven, CT

Low statistical power in neuroimaging often undermines research in the field, leading to missed effects, wasted resources, and reduced reproducibility. Performing power analyses during the study design phase is extremely important, but often prohibitively difficult due to a lack of analytical solutions tailored to complex fMRI statistical procedures, the challenge of defining expected effect sizes a priori, and high computational costs. We present \texttt{PRISME} (Power Reproducibility and Inference for Statistical Method Evaluation), a MATLAB toolbox for neuroimaging power calculations. \texttt{PRISME} provides a computational framework for comparing multiple statistical inference methods, enabling researchers to select the correct sample sizes for their experiments and find optimal methods for their experimental data analysis and integrate new ones for benchmarking through a flexible algorithm and architecture. Unlike previous empirical power approaches to calculate power, the toolbox supports diverse neuroimaging data types, including both voxel-based activation and functional connectivity analyses, and arbitrarily complex inferential methods due to a flexible nonparametric algorithm and unified data representations. Furthermore, \texttt{PRISME} supports multiple test types, such as association and difference tests with behavioral and clinical measures. Finally, \texttt{PRISME}'s 25× increase in speed from algorithmic optimizations compared with procedures in existing scripts enables larger-scale never before seen power calculations which were previously computationally infeasible. Overall, \texttt{PRISME} is the first method- and data-type-agnostic power calculator for neuroimaging, providing an accessible solution for power analysis across diverse study designs.

Topic Area: METHODS: Neuroimaging