A Guide to Literature Informed Decisions in the Design of Real Time fMRI Neurofeedback Studies: A Systematic Review

Figures

• 

  Figure 1. Diagram of a generic neurofeedback experiment with study design elements highlighted. (A) Functional MRI data is collected from a patient, focusing either on a single region of interest (ROI), multiple ROIs, or the whole brain. (B) Data from the subject is processed in real-time, usually employing standard neuroimaging software packages. Feedback is calculated, sometimes based on ROI signal change, correlation between several ROIs, or using a classification algorithm. This may require an offline fMRI scan prior to neurofeedback training to establish baseline activity or to train the classifier. (C) The feedback value is transformed and given to the patient in the scanner, often via a continuous visual graph displayed on a screen. (A) This allows the participant to react and change their neural response, starting the feedback cycle over.

• 

  Figure 2. Diagram of literature search procedures across the primary and update literature search dates based on PRISMA 2009 Flow Diagram (Moher et al., 2009). fMRI, functional magnetic resonance imaging; rt-fMRI-NF, realtime functional magnetic resonance imaging neurofeedback training.

• 

  Figure 3. (A) Histogram of publication dates for all empirical rt-fMRI neurofeedback studies. (B) rt-fMRI neurofeedback studies categorized by task type. NA studies had no stimuli or instructions that fell into a particular task domain. Numbers indicate the number of studies in each category. (C) rt-fMRI neurofeedback studies by type of sample used in the study. Numbers indicate the number of studies in each category.

• 

  Figure 4. (A) rt-fMRI neurofeedback studies categorized by signal/algorithm type. Classification includes SVM, DecNef, MVPA etc. Numbers indicate the number of studies in each category. (B) rt-fMRI neurofeedback studies that employed an ROI signal approach by region used for that ROI. Numbers indicate the number of studies in each category. ROI, region of interest; ACC/PCC, anterior/posterior cingulate cortex; OFC, orbitofrontal cortex.

• 

  Figure 5. (A) rt-fMRI neurofeedback studies categorized by direction subjects were instructed to regulate Numbers indicate the number of studies in each category. (B) Stacked bar char of rt-fMRI neurofeedback studies using the ROI approach broken down by direction of regulation and ROI. Numbers indicate the number of studies in each category. ROI, region of interest; mPFC, medial prefrontal cortex; ACC/PCC, anterior/posterior cingulate cortex; OFC, orbitofrontal cortex; SMA, supplementary motor area.

• 

  Figure 6. (A) rt-fMRI neurofeedback studies categorized by the way feedback was displayed to subjects. Numbers indicate the number of studies in each category. (B) rt-fMRI neurofeedback studies by feedback timing. Numbers indicate the number of studies in each category. (C) rt-fMRI neurofeedback studies by nature of the instructions for regulating brain activity. Numbers indicate the number of studies in each category.

• 

  Figure 7. (A) Histogram of sample sizes used in rt-fMRI neurofeedback studies. Red line indicates median, blue line indicates mean. (B) rt-fMRI neurofeedback studies by nature of the control group used to compared the active neurofeedback condition. Numbers indicate the number of studies in each category. ROI, Region of interest. (C) Histogram of number of runs used in rt-fMRI neurofeedback training sessions. Red line indicates median, blue line indicates mean. (D) Histogram of number of sessions used in rt-fMRI neurofeedback training protocols. Red line indicates median, blue line indicates mean.

• 

  Figure 8. (A) rt-fMRI neurofeedback studies by nature of the outcome measure used to evaluate neurofeedback effectiveness. Numbers indicate the number of studies in each category. (B) Histogram of follow-up times used in rt-fMRI neurofeedback studies with follow-up sessions. Red line indicates median, blue line indicates mean.