Using multimodal neuroimaging to predict responses to electrical brain stimulation in disorders of consciousness

It is well accepted that some patients in a vegetative state (VS) retain a much greater deal of cognitive function than initially thought, and are simply unable to produce external purposeful responses.1 We have demonstrated that the lack of voluntary control these patients exhibit can be explained by specific impairments in structural connectivity within the motor system.2 As part of a recently awarded three-year grant (Medical Research Council [MRC]; 2017-2020), the Principal Investigator, and primary supervisor for this PhD project (DF-E), is exploring the potential for non-invasive brain stimulation (transcranial direct current stimulation; tDCS), coupled with passive mobilisation, to modulate the dynamics of the motor system and subsequently increase motor responsiveness. By using concurrent functional magnetic resonance imaging (fMRI) and tDCS, this work will characterise the neural mechanisms underlying the behavioural effects of tDCS in healthy individuals. Importantly, data from both healthy studies3 and preliminary work in VS patients and related disorders of consciousness (e.g., 4) suggest heterogeneity in individual responses to tDCS. One specific aim of the grant, which will be the focus of the PhD project proposed here, is thus to understand and characterise individual variations in behavioural and neural effects of tDCS, and to identify biomarkers that can predict whether an individual will be likely to respond to a particular stimulation protocol.

In collaboration with a postdoctoral researcher, the student will first test healthy participants with concurrent tDCS-functional magnetic resonance imaging (fMRI) to measure the neural effects of different stimulation/passive mobilisation protocols both under the stimulating electrodes as well as in remote brain regions. All stimulation protocols will be designed to be feasible in VS. The student will also use motion tracking to record changes in behavioural responses during simple motor tasks. The experiments will involve different theoretically motivated target regions, and dosage levels (i.e., number of sessions). To fully characterise individual differences in neural architecture, the student will focus on advanced analyses of structural and functional MRI data (i.e., cortical thickness, diffusion tensor tractography, and functional and effective connectivity at rest). The student will assess how differences in the neuroanatomy and function of the motor network modulate individual responses to stimulation. Specifically, they will pool metrics extracted from the above MRI modalities and test the power for different classification techniques (i.e., Support Vector Machines, and Deep Learning). Additionally, the student will explore the construction of Decision Tree classifiers. These provide explicit information about why the algorithm came to a particular classification decision and, thus, can have higher value in medical contexts, where justification can be very important.

In parallel, as part of a collaboration agreement between the primary supervisor (DF-E) and the Wellington Hospital, the student will test their methods in a cohort of patients with a disorder of consciousness. The experimental design and analysis pipeline will mirror that for the experiments in healthy volunteers above. However, clinical data (i.e., clinical history as well as responses in diagnostic behavioural scales) will also be pooled with the multimodal MRI data for testing classification methods.