Electronic Thesis and Dissertation Repository

Thesis Format

Integrated Article


Doctor of Philosophy




Soddu, Andrea

2nd Supervisor

Valluri, Sreeram



Although many technical advancements have been made, neuroscientists still struggle to explain the underlying behaviour of how brain regions communicate with each other to form large-scale functional networks. functional Magnetic Resonance Imaging (fMRI) has been commonly used to investigate changes between brain regions over time using the Blood Oxygen Level Dependent (BOLD) signal.

The goal of this thesis is to show the applicability of novel techniques and tools, such as the generalized Ising model (GIM) and the independent component graph analysis (GraphICA), to obtain information on the functional connectivity of populations with altered perception of consciousness. The GIM was used to model brain activity in healthy brains during various stages of consciousness, as induced by an anesthetic agent, propofol, in the auditory paradigm. GraphICA, a tool that combines ICA and graph theory was used to investigate the functional connectivity of resting state networks (RSNs) in patients with altered perception caused by tinnitus and in patients with altered states of consciousness caused by severe brain injury. For the tinnitus patients, we examined whether a correlation exists between tinnitus behavioural scores and functional brain connectivity of RSNs. Moreover, for the severely brain injured patients, a multimodal neuroimaging approach with hybrid FDG-PET/MRI was implemented to study the functional connectivity changes of the RSNs.

The GIM simulated with an external field was able to model the brain activity at different levels of consciousness under naturalistic stimulation, at a temperature in the super critical regime. Further, a strong correlation was observed between tinnitus distress and the connectivity pattern within and between the right executive control network and the other RSNs. This suggests that tinnitus distress is the consequence of a hyperactive attention condition. A variability was observed in the appearance and temporal/spatial patterns of RSNs among the two resting state fMRI acquisitions acquired within the same scanning session of the severely brain injured patients. This suggests the need for new strategies to be developed in order to pick the best RSN from each acquisition. Overall, this work demonstrated that the GIM and GraphICA were promising tools to investigate brain activity of populations with altered perception of consciousness and in future can be extended to investigate different neurological populations.

Summary for Lay Audience

The human brain is one of the largest and most complex organs. It acts as the commanding center of the nervous system. It is made up of more than a hundred billion neurons that work together to successfully generate and transfer information. Despite recent advancements in neuroscience, researchers still struggle to explain how the brain functions in order to effectively complete specific tasks.

Our research is to map out the differences in functional connectivity in altered states of consciousness, arising either by the intake of an anesthetic agent or due to pathological conditions. In the first project, we successfully modelled the brain activity of healthy subjects while they were listening to an audio clip from a movie, in awake, mildly sedated, deeply sedated and recovered back, due to anesthesia. Further, synchronization among subjects’ brain regions responsible for perception was found only when the individuals were listening to the movie. In future, this simple model will permit us to see whether a non-responsive patient has the same neural experience of a healthy individual.

In the second project, a study focussed on tinnitus (ringing in ears without any sound) patients, we found that distress resulted due to hyper-activity of brain regions in charge of executive functioning. This concludes that tinnitus distress is mainly caused by hyper attention towards internally generated auditory sensations, “auditory hallucinations”. In our final project on patients with disorders of consciousness, we observed a variability between MRI scans acquired within a temporal gap of 30 min. This suggests a necessity to perform multiple scans in the same session to reduce misdiagnosis.