Abstract
The brain has an ordered structure of functional connections within brain networks, but this order becomes disrupted in Alzheimer’s disease (AD). The default mode network (DMN) and its central hub, the Posterior Cingulate Cortex (PCC) are particularly affected. Disruption to the functional relationships results in AD memory symptoms. Electroencephalography (EEG) biofeedback analyses a subject’s EEG, and displays an aspect of that signal back to them. That subject can then take control over that aspect of their EEG signal. Biofeedback can be used to alter the functional relationships in brain networks and could restore the abnormal functional connectivity relationships seen in AD, and in doing so improve memory symptoms. The aim of this thesis is to examine whether source-localised EEG biofeedback of the PCC could alter functional connectivity in the earliest stages of AD and improve memory outcomes.
First, a pilot study in 10 people with amnestic type Mild Cognitive Impairment (aMCI) was conducted. The protocol was designed to up-train the power of theta and alpha, source localised to the PCC (broadband feedback). Participants completed 15 sessions of 40 minutes of biofeedback training. Their memory was assessed using the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS), which has two forms (A and B). Participants were tested using form A prior to training, and form B after training to reduce repeated testing effects. There was a significant improvement in both the RBANS Immediate Memory index (mean improvement 12.4 points, p=0.011) and the Delayed Memory index (mean improvement 7.0 points, p=0.016).
Following on from the pilot study, a three armed randomised controlled trial was designed to:
1) Assess alpha band only source localised feedback (narrowband) against broadband feedback
2) Assess both biofeedback paradigms against a placebo control
3) Test biofeedback training in a more general population of adults with memory deficits
Volunteers were screened using the RBANS. Those who scored below 90 on the immediate memory index were randomised to receive 15 sessions of either broadband feedback, narrowband feedback or placebo feedback. Memory was assessed using the RBANS.
53 people completed all 15 sessions of biofeedback training and were included in the analysis. No effect on memory was found (Randomisation*Time interaction = 1.162, p=0.388). A significant increase in the power of the alpha band in the narrowband feedback group was identified (β=0.0323, p=0.003), indicating that the training had an effect on the targeted frequency in this group; however no change in the targeted frequency in the broadband feedback group was detected. No significant impact on DMN connectivity was found. A significant difference was found in the difficulty of tasks in the RBANS immediate memory index, form A was more difficult than form B.
This particular kind of source localised biofeedback training does not appear to be effective at improving memory outcomes in the short term. In addition, a significant form effect was discovered for the RBANS, which entirely explained the result of the pilot study. Future studies of biofeedback should use a counterbalanced design when using alternate forms of neuropsychological tests.