Exploring diffusion imaging as a predictor of anxiety disorders

• Main Author: Liauw, Claudia Yong Tong
• Other Authors: -
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Science::Biological sciences::Human anatomy and physiology::Neurobiology
• Online Access: https://hdl.handle.net/10356/166643
• Institution: Nanyang Technological University

Anxiety disorders are the most common mental disorders, but diagnoses are based on subjective symptoms. Thus, there is active interest to find objective biomarkers of anxiety such as in neuroimaging. Some diffusion imaging markers have been found, but most studies have small sample sizes, rendering effect sizes too small for individual predictions. This study explored data from the Healthy Brain Network (HBN), one of the largest youth datasets available. It aimed to replicate findings of tracts associated with generalised anxiety disorder (GAD) or social anxiety disorder (SAD) in the literature using statistical analysis and use machine learning to model diffusion imaging data in order to predict GAD or SAD. Analyses of a dataset of 318 individuals from the HBN dataset could not replicate existing findings in the literature and the tract-based diffusion imaging markers were unable to predict GAD nor SAD. Machine learning models showed significant prediction of SAD from demographics data, but the prediction score was not high. The results suggest that tract-based properties from diffusion imaging may need to be augmented with other modalities such as function and structure to capture individual differences in anxiety.