PhD Studentship: Generative AI for Synthetic Microscopy Image of Healthy Oral Mucosa

Project Outline

Understanding the structure of healthy oral mucosa is essential for studying function and physiological changes over time. It also provides a foundation for identifying pathological changes, supporting the discovery of disease markers and therapeutic targets. However, obtaining healthy oral mucosa samples is challenging, as biopsies from healthy tissues are rarely taken. Most studies rely on surrogate tissues from lesion margins or benign conditions, which may not fully reflect healthy tissue. This limits our ability to build high-quality reference datasets. With advances in artificial intelligence (AI), it is now possible to generate realistic synthetic images, offering an ethical and scalable way to expand healthy tissue collections and support future research. This project will use generative AI to establish a large reference microscopy images dataset of healthy oral mucosa.

The project aims to develop AI models that replicate the appearance and structure of healthy oral mucosa. Objectives include:
1. Training generative models to produce realistic, high-resolution microscopy images.
2. Ensuring no synthetic image is identical to real patient data.
3. Evaluating image quality, diversity, and biological relevance using standard metrics and expert review.

Anonymised digital images from tissues in biobanks will be used to train generative models on university computing resources, including 200+ NVIDIA A100 GPUs and group workstations. Image quality will be assessed using quantitative metrics and clinical expert qualitative review. Privacy safeguards will be built into the generation process.

This multidisciplinary project will deliver deployable models, reproducible methods, and, where allowed, shareable datasets. The student will gain training in deep learning, AI, image analysis, microscopy, digital pathology, and oral mucosa biology. Generic skills training is provided through the Postgraduate School.

Supervisors: Dr Qingjie Meng (m.qingjie@bham.ac.uk) and Dr Rasha Abu-Eid.

Funding notes:

This is a PhD studentship with the Midlands Integrated Biosciences Training Partnership, funded by BBSRC and in partnership with the University of Warwick, Aston University, Harper Adams University, Coventry University, and the University of Leicester.

For more details please visit: https://warwick.ac.uk/fac/cross_fac/mibtp/ or https://www.birmingham.ac.uk/about/college-of-life-and-environmental-sciences/midlands-integrative-biosciences-training-partnership

To apply:

To apply, please follow this link, make an account, and submit an application via the university online admissions portal (via the above ‘Apply’ button). This link is unique to the MIBTP programme; please do not use any other link to apply to this project or your application may be rejected.

Advert information