PhD Studentship: Development of AI Strategies for Accurate Tumour Staging in Multisite Cancer Imaging

Overview: This PhD studentship (starting October 2025), open to UK and overseas candidates, offers a unique opportunity to transform cancer diagnosis and treatment by significantly improving the use of positron emission tomography (PET) imaging in modern medicine. PET scans help doctors diagnose cancer, understand how far it has spread, and monitor how well treatments are working. However, variations in scan quality across different hospitals or imaging sites can make it difficult to get consistently accurate results. This project will develop new methodologies that ensure PET scans are accurate and reliable, regardless of where and how they are acquired. The PhD candidate will be working between two academic research centres (London South Bank University and University College London) with substantive industry involvement from General Electric Healthcare and Alliance Medical Ltd, whose imaging network covers 38 sites across England.

Background: Lung cancer is one of the deadliest cancers, and PET scans play a crucial role in detecting, staging and managing cancer by revealing the metabolic activity of tumours with details and insights that other imaging methods cannot match (for more details see this review https://doi.org/10.3390/cancers6041821). However, PET images are often blurry or noisy, especially for small tumours, and scan quality varies substantially across hospitals due to differences in imaging technology. This inconsistency makes it difficult to compare scans over time or between institutions.

Research Problem: The main challenge this project aims to address is the inconsistency in PET scan quality in lung and lymphoma cancers across different imaging sites and hospitals. As technology evolves, newer scanners produce higher-quality images, but not all hospitals have access to the latest equipment. This variation can lead to significant differences in how cancer is diagnosed and monitored, and often the better images are either downgraded or left unused.

Research Goals: This project seeks to standardise PET imaging and scoring ( https://radiopaedia.org/articles/herder-risk-model, https://radiopaedia.org/articles/deauville-five-point-scale) across sites without downgrading better-quality scans. Instead, it will enhance imaging using advanced reconstruction techniques and AI, optimising settings for modern scanners while compensating for differences in technology, patient size, and radiation dose. This will improve image clarity, aiding doctors in accurate interpretation.

The student will work with diverse PET scan data from hospitals across England, tackling real-world cancer imaging challenges. Possible research focus includes development of mathematical and physical models for image reconstruction and novel image analytics based on AI-driven enhancements. Ideal candidates should have strong programming skills and a background either in computer science, engineering, machine learning, or medical sciences, with a passion for improving cancer care.

Please contact Dr. Pawel Markiewicz for more details at pawel.markiewicz@lsbu.ac.uk or p.markiewicz@ucl.ac.uk.

Advert information