Year: 2020
Type: dissertation
Abstract: The diagnosis of psychiatric disorders often relies on clinician’s interviews and self-reported questionnaires. These techniques suffer from inter-observer variability and are often affected by communication difficulties between clinicians and patients. Analysis of visual scanning behaviour (VSB) is a method that can help to objectively characterise patients with psychiatric disorders. The main goals of this thesis are to a) develop novel techniques to analyse VSB and b) to develop novel experimental paradigms that are based on the analysis of VSB to assess cognitive processes. Data driven methodologies based on recurrent neural networks (RNN) were developed to learn visual scanning sequences of patients with mood disorders and patients with Alzheimer’s disease (AD). The method that was developed for patients with mood disorders is based on the analysis of VSB on human faces and can differentiate between patients with bipolar or unipolar depression with an area under the curve (AUC) of 0.879 (a gain of 0.276 relative to expert systems). The method that was developed to detect emotional blunting in AD patients could learn the individual’s “style of scanning” and can differentiate between apathetic and non-apathetic patients with AUC of 0.814 (0.646 for an expert system). Novel experimental paradigms were developed to assess the potential for conversion to AD in mild cognitive impairment (MCI) patients and a surrogate marker for the detection of structural damage to white matter tracts in patients with post-concussion syndrome (PCS). The paradigm for the assessment of MCI patients is based on measurements of novelty preference (NP) where repeated images appear after short delays (≤12.5 seconds). For short delays, NP is dependent on intact familiarity-based recognition memory. The marker for structural damage to white matter tracts in PCS patients is based on a novel visual scanning version of the matching familiar figures test (MFFT). VSB measures during the MFFT test were correlated to the microstructural integrity of the corpus callosum (r=-0.281, p=0.0141). In summary, the work in this thesis shows that by using novel paradigms and novel analysis techniques, VSB can provide valuable insights that can support the diagnosis and treatment of patients with psychiatric disorders.
Source: TSpace (University of Toronto)
Author: Jonathan Man Kit Chung
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W3154902457