Effects of fMRI neurofeedback of right inferior frontal cortex on inhibitory brain activation in children with ADHD

Abstract

We aimed to replicate previous effects of functional magnetic resonance imaging neurofeedback (fMRI-NF) in right inferior frontal cortex (rIFC) on IFC activation during a Stop Task in a larger group of boys with attention-deficit/hyperactivity disorder (ADHD). The present double-blind, randomized controlled trial tested the effects of 15 runs of active versus sham fMRI-NF of rIFC on performance and activation associated with successful and failed inhibition versus Go trials during a tracking Stop task in 88 boys with ADHD (44 active; 44 sham), controlling for age and medication status. No significant group-by-time interaction effects were observed for performance or brain activation during the successful stop trials, and post hoc analysis showed very low numbers of active fMRI-NF learners. Nevertheless, during error monitoring, there was a significant group-by-time interaction effect on post-error reaction time slowing and in left IFC activation, which were both increased after active compared to sham fMRI-NF. The findings are in line with our previous observation of left IFC upregulation after fMRI-NF of rIFC relative to active fMRI-NF of parahippocampal gyrus. This highlights the potentially wider regional effects that fMRI-NF of a particular self-control target region has on other self-regulatory regions in ADHD. This article is part of the theme issue 'Neurofeedback: new territories and neurocognitive mechanisms of endogenous neuromodulation'.

Keywords: ADHD; children; fMRI neurofeedback; inhibition; stop-signal task.

Figure 1.

The overall trial design, fMRI-NF set-up and rIFC activation across runs. (a) This study was part of a larger trial assessing the efficacy of fMRI-NF for children with ADHD that consisted of seven visits including four fMRI scan sessions. The fMRI Stop task occurred during the scan visit A and D before the first fMRI-NF run and after a transfer run. (b) This diagram depicts the fMRI-NF architecture. Each fMRI-NF run consisted of seven rest (R) and six self-regulation (S) blocks. During the S block, (b(i)) the participant brain scan images were acquired in the MRI scanner, (b(ii)) reconstructed in the MRI station, (b(iii)) pre-processed using the Analysis of Functional NeuroImages (AFNI) in a remote server to compute (b(iv)) fMRI-NF signal that enabled (b(v)) control of the rocketeer that was visually fed back to the participant. (c) The average BOLD activation extracted from the rIFC cluster for the sham and the active groups is depicted across 15 runs. (d) The same signal from individuals in the active group divided into those who were judged as learners and non-learners.

Figure 2.

Group × time Interaction of post error response time slowing (PERTS). Simple-effect analyses showed a significantly higher PERTS in the sham versus active group at pre-treatment (p = 0.043) and a decrease of PERTS in the sham group from pre- to post-treatment (p = 0.007). Error bars indicate the standard error of the mean. Significant thresholds: *p < 0.05, **p < 0.01.

Figure 3.

Within-group and group x time interaction brain activation clusters. (a) Within-group activation clusters, uncorrected at peak p < 0.001 and family-wise corrected cluster level p _FWE < 0.05, associated with (i) successful Stop versus Go trials, and (ii) failed Stop versus Go trials in the active and sham group at pre- and post-treatment scans. The Go trials were modelled as an implicit baseline. (b) Group x time interaction revealed significant clusters at the left IFC after small volume correction at the region of left IFC combined opercular and triangular parts at uncorrected peak p < 0.001.