IntroductionFunctional magnetic resonance imaging (fMRI) often involves long scanning durations to ensure the associated brain activity can be detected. However, excessive experimentation can lead to many undesirable effects, such as from learning and/or fatigue effects, as well as undue discomfort for the subject, which can lead to motion artifact and loss of sustained attention on task. Overly long experimentation ironically can thus have a detrimental effect on signal quality and accurate voxel activation detection. Here, we propose a method of dynamic experimentation with real-time fMRI using a novel statistically-driven approach to fMRI analytics. This new approach to experimental design invokes early stopping when sufficient statistical evidence for assessing the task-related activation is observed.MethodsVoxel-level sequential probability ratio test (SPRT) statistics based on general linear models (GLM) were implemented on fMRI scans of a mathematical 1-back task from 25 subjects, 14 healthy controls and 11 subjects born extremely preterm. This approach is based on likelihood ratios and allows for systematic early stopping based on statistical error thresholds being satisfied. We explored voxel-level serial covariance estimation in real-time using the “sandwich” estimator. We adopted a two-stage estimation approach that allows for the hypothesis tests to be formulated in terms of t-statistic scale, which enhances interpretability. Scan data was collected using a dynamic feedback system that allowed for adaptive experimentation. Numerical parallelization was employed to facilitate completion of computations involving a new scan within every repetition time (TR).ResultsSPRT analytics demonstrate the feasibility and efficiency gains of automated early stopping, while performing comparably in activation detection with full protocols analyzed through standard fMRI software. Dynamic stopping of stimulus administration was achieved in all subjects with typical time savings between 33 - 66% (4 – 8 minutes on a 12 minute scan).ConclusionA systematic statistical approach for early stopping with real-time fMRI experimentation has been implemented. This allows for great savings in scan times, while still eliciting comparable activation patterns as full protocols. This dynamic approach has promise for reducing subject burden and fatigue effects.