Power and sample size calculations for fMRI studies based on the prevalence of active peaks

Abstract

Highlights• The manuscript presents a method to calculate sample sizes for fMRI experiments • The power analysis is based on the estimation of the mixture distribution of null and active peaks • The methodology is validated with simulated and real data. AbstractMounting evidence over the last few years suggest that published neuroscience research suffer from low power, and especially for published fMRI experiments. Not only does low power decrease the chance of detecting a true effect, it also reduces the chance that a statistically significant result indicates a true effect (Ioannidis, 2005). Put another way, findings with the least power will be the least reproducible, and thus a (prospective) power analysis is a critical component of any paper. In this work we present a simple way to characterize the spatial signal in a fMRI study with just two parameters, and a direct way to estimate these two parameters based on an existing study. Specifically, using just (1) the proportion of the brain activated and (2) the average effect size in activated brain regions, we can produce closed form power calculations for given sample size, brain volume and smoothness. This procedure allows one to minimize the cost of an fMRI experiment, while preserving a predefined statistical power. The method is evaluated and illustrated using simulations and real neuroimaging data from the Human Connectome Project. The procedures presented in this paper are made publicly available in an online web-based toolbox available at www.neuropowertools.org.

Power, positive predictive value, and sample size calculations for random field theory-based fMRI inference

“…Specically, in terms of its power function, a region of interest-based approach corresponds to uncorrected inference at the voxel level, i.e., a power evaluation for a one-sample T -test, with the dierence that in typical region of interest-based approaches, voxel height statistics are spatially averaged over a set of voxels. Another power calculation framework that has recently been popularized is the approach of Durnez et al (2016). This framework rests on a testing procedure that considers local maxima of voxel height statistics above a threshold.…”

“…This framework rests on a testing procedure that considers local maxima of voxel height statistics above a threshold. Under the model by Durnez et al (2016), these local maxima are thought to be the outcome of a mixture distribution, comprising realizations of a null hypothesis exponential distribution and an alternative hypothesis Gaussian distribution. While the test procedure itself is not explicitly described, the apparent idea is to reject the null hypothesis of no activation at the location of the local maximum based on a set of arbitrary selected critical values (Durnez et al, 2016, Section 3.3 (1994), while newer results for T -and F -elds are available (e.g., Cao, 1999).…”

“…First and foremost, most previously proposed frameworks are not well aligned with the theory of RFT-based fMRI inference (e.g., Desmond and Glover, 2002;Mumford and Nichols, 2008;Durnez et al, 2016), rendering them non-applicable for the most commonly employed forms of fMRI inference. Second, the framework previously proposed by Hayasaka et al (2007) and Joyce and Hayasaka (2012) that is aligned with the theory of RFT-based fMRI inference only addresses voxel-level and not cluster-level inference.…”

A methodological assessment of studies that use voxel-based morphometry to study neural changes in tinnitus patients

“…Sample size calculations for neuroimaging studies are challenging and their omission is common. However, in recent years new tools have been developed to facilitate this important step in study design (Mumford and Nichols 2008; Joyce and Hayasaka 2012; Durnez et al 2016), which need not always require pilot data. In fact, access to prior study data or statistical maps to facilitate power analysis is a benefit of increased data sharing.…”

A Hitchhiker's Guide to Functional Magnetic Resonance Imaging

“…In order to employ these tools, information about the mean activation, the variance, the Type I error rate, and the sample size must be provided (Mumford, 2012). The power calculation should use either the statistical images (t/F maps generated by simple study designs) from pilot studies (PowerMap software; Joyce and Hayasaka, 2012), the estimated parameters in specific regions-of-interest (fMRIPower tool) (Mumford and Nichols, 2008) or the prevalence of active peaks (NeuroPower; Durnez et al, 2016). …”