About the Project About the project Magnetic resonance imaging (MRI) provides detailed information about brain structure and has enormous potential
to improve our understanding of brain development and mental health. Realising this potential requires accurate and scalable methods for identifying brain anatomy and deriving meaningful measurements from large imaging datasets. Manual analysis is time-consuming and subject to variability, while established automated tools such as FreeSurfer can take hours to process a single scan, are sensitive to image artefacts and can be difficult to adapt to new imaging modalities. Recent advances in artificial intelligence (AI), particularly deep learning, offer an opportunity to develop faster, more robust and adaptable approaches. This project will use the Avon Longitudinal Study of Parents and Children (ALSPAC), a world-leading birth cohort with more than three decades of biological, genetic, psychological and health data. ALSPAC includes three neuroimaging sub-studies with multimodal MRI, providing a unique setting in which to develop AI methods and examine how improved brain measurements can advance population health research. Aims and objectives The project aims to develop and apply an AI-driven pipeline for population neuroimaging. The student will: Develop and validate state-of-the-art and novel deep-learning methods for segmenting brain structures from MRI. Generate high-quality brain phenotypes, such as cortical thickness, surface area, subcortical volumes and global tissue volumes, and investigate additional measurements enabled by the new methods. Develop and evaluate strategies to improve model robustness and generalisability across datasets and imaging protocols, including uncertainty estimation and automated quality control, and benchmark performance and computational efficiency against established approaches such as FreeSurfer. Conduct a focused proof-of-principle analysis relating selected brain phenotypes to ALSPAC's longitudinal, genetic, behavioural and psychological data to investigate brain development and mental health. The project has a clear overall framework while allowing the student to shape its direction. Depending on their interests, they may focus on algorithmic innovation, multimodal MRI, automated quality control, model generalisability, particular biological or mental health questions, or the development of reusable tools for the wider neuroimaging community. Methods The student will begin by reviewing existing neuroimaging pipelines and establishing benchmark results using FreeSurfer. They will implement state-of-the-art segmentation approaches, potentially including nnU-Net, attention-based networks and transformer architectures. Models will initially be trained or evaluated using suitable public datasets with manual annotations, then adapted and fine-tuned for ALSPAC MRI data using transfer-learning or domain-adaptation strategies where appropriate. Performance will be assessed using quantitative segmentation metrics, expert-informed quality control, robustness to artefacts and variation in acquisition, and computational efficiency. Beyond segmentation accuracy, the student will investigate strategies to improve robustness and generalisability across datasets and imaging protocols, potentially including uncertainty estimation, automated quality control and domain adaptation. Generalisability will be evaluated across the ALSPAC neuroimaging sub-studies and suitable public datasets acquired using different protocols. Subject to data access and relevant approvals, selected models and derived brain phenotypes may also be evaluated in an independent cohort such as Generation R, providing an opportunity to investigate cross-cohort transportability and domain shift. The student will compare the reliability and biological utility of the resulting phenotypes with existing FreeSurfer-derived measures. Finally, they will relate selected brain phenotypes to ALSPAC's longitudinal, genetic, behavioural and psychological data in a focused proof-of-principle analysis of brain development and mental health. The project will emphasise reproducible workflows, transparent reporting and careful evaluation of limitations and generalisability. How to apply Before applying, please check the entry requirements . Submit your application via the University of Bristol portal . Search for and then select ‘Population Health Sciences (PhD)’. Click ‘Apply’ for September 2027 start. You must provide the following in your application: A CV Two referees (add to the Reference section of your application form) A personal statement highlighting your motivation for applying for this PhD, and any relevant experience and/or skills (maximum of one side of A4, font size 11). We are keen to support applicants from minority and under-represented backgrounds (based on protected characteristics ) and those who have experienced other challenges or disadvantages. Please use your personal statement to ensure we can take these factors into account. Certificates and transcripts Enter the project title and the supervisor names as they appear on this advert in the Research section. A research statement/proposal is not required. You should upload a blank document in this section. In the funding section, choose ‘University of Bristol Scholarship’. Incomplete applications will not be considered . Read the list above carefully and ensure you have uploaded all required documents by the deadline. If you have yet to complete your degree, please supply an interim transcript where possible. You can find more information on how to prepare your application and what we are looking for at the “PhD Application Workshop” which will be held on Teams on Monday 23rd November at 2pm (GMT): PhD Application Workshop. Applications will be shortlisted in January 2027 with interviews in February 2027. The anticipated start date for this scholarship is September 2027. Please visit our website for more information on PhD opportunities in the Faculty of Health and Life Sciences. Further enquiries Project-specific queries: yi.yin@bristol.ac.uk Application queries: uob-pgradmissions@bristol.ac.uk Funding Notes The University of Bristol is one of the most popular and successful universities in the UK, ranked 51st globally (QS 2026), top five in the UK for research (THE analysis of REF 2021), and 6th most targeted by UK employers (High Fliers 2026). This is a four-year scholarship covering tuition fees, a UKRI-rate stipend (£21,805 in 2026/27, updated annually), and £2,100 annual research costs. The scholarship is available to UK and International applicants. International candidates (including EU) will be required to cover the cost of their student visa, healthcare surcharge and other costs of moving to the UK. References https://doi.org/10.1016/j.neuroimage.2012.01.021 https://doi.org/10.12688/wellcomeopenres.16060.1 https://doi.org/10.1038/s41592-020-01008-z


