This thesis investigates neuroanatomical variation in relation to autism and attention deficit hyperactivity disorder (ADHD), using a variety of structural neuroimaging techniques. Autism and ADHD are heterogeneous neurodevelopmental conditions with complex underlying biology that is still poorly understood, despite efforts to characterise these conditions at the neuroanatomical level. Though autism and ADHD have overlapping presentation and sex- biased prevalence, they are rarely studied together, and sex differences are often overlooked. Numerous challenges in this field are still hampering progress, including variable image quality, methods for combining multi-site datasets, heterogeneity of clinical populations, and lack of clinical applications. The autistic community is also rarely involved in biomedical research efforts, raising further questions about translational potential and who this research is serving.

Chapter 1 of this thesis reviews the literature and state of the field of structural brain alterations and development in autism and ADHD, and relevant neuroimaging methods and techniques used to study them. It also identifies several challenges and opportunities for future research that need to be addressed in order to advance this field of research.

Chapter 2 introduces a new quality control (QC) method and tool, FSQC, for the quick and efficient rating of cortical reconstructions as well as raw image quality. It then explores how variations in magnetic resonance imaging (MRI) quality can impact and bias downstream analyses in neurodevelopmental imaging. FSQC was validated against a range of different automated and manual QC methods, which were shown to capture both distinct and overlapping relationships with cortical morphometry. Applying varying levels of QC stringency was shown to attenuate, but not eliminate, relationships between quality and cortical neuroanatomy. Importantly, failing to account for quality had subtle but significant impacts on apparent case-control differences related to autism, and thus could be an important confound in studies of neurodevelopmental and psychiatric conditions.

Chapters 3 and 4 investigate structural variation in global and cortical neuroanatomy, and divergences from typical brain development, in autism and ADHD. Population modelling, sometimes referred to as normative modelling, provides a unified framework for studying age- and sex-specific divergences in brain development. After applying the QC methods described in Chapter 2 to a large, multi-site neurodevelopmental dataset, a population modelling approach was used, leveraging existing models of typical brain development across the lifespan. Largely distinct alterations were observed in autism and ADHD, with some overlap. Autistic individuals, on average, presented with significant increases in cortical thickness and volume, localised to the superior temporal gyrus, whereas ADHD was associated with global increases in cortical thickness, and decreases in cortical volume and surface area. Distinct differences were also observed in a subset of individuals with a dual diagnosis of autism and ADHD, who presented with widespread increases in cortical thickness. Chapter 4 explored sex differences in these results, identifying sex-specific alterations in autism but not ADHD, and highlighting the importance of taking sex into account in studies of neurodevelopmental conditions.

Chapters 5 extends this work by attempting to parse the heterogeneity of both conditions along two dimensions: age and the presence of clinically relevant subgroups within or across conditions. An age sliding-window analysis revealed the greatest group differences in the first few years of life for autism, but in late adolescence for ADHD. Multiple subgrouping approaches were used, including one which identified subgroups based on developmental trajectories. The number of subgroups identified varied by approach from 2-4, with some significant clinical differences between groups observed. Subgroup identity of participants varied between approaches, with some overlap.

Chapters 6 presents the results of a community engagement focus group and survey conducted to gain insight into the autistic community’s thoughts, perceptions, and priorities in relation to neuroimaging research. Focus group participants helped to design the survey, which was completed by over 400 participants (of which approximately 90% were autistic individuals, and 10% were non-autistic parents of autistic children). Perceptions about neuroimaging research in autism were overall positive, although many participants had significant concerns, primarily relating to ethics, secondary data use and misuse, practical utility and clinical translation of results, and the potential use of such research to attempt to cure or eradicate autism. Some of the top research priorities of the autism community included developing earlier and more effective and objective diagnostic tools, developing and improving individualised supports for autistic individuals, and improving our basic understanding and knowledge of the underlying neurobiology and differences related to autism.

Finally, Chapters 7 summarises these results, contextualises them in light of current knowledge, and identifies convergent themes and findings. Future directions and potential routes for clinical translation are also discussed.