Automated Longitudinal Segmentation of Glioblastoma on the MR-Linac active

Lead: Daryl Wilding-McBride

Glioblastoma is the most aggressive primary brain tumour in adults, with a median survival of around 15 months despite surgery, radiotherapy, and chemotherapy. Tumours are diffuse, morphologically heterogeneous, and can shift substantially during the treatment course, yet standard radiotherapy planning treats them as static. The MR-Linac addresses this by acquiring MRI at every treatment fraction, creating a longitudinal imaging record unique among radiotherapy modalities. In current clinical practice this data is largely underused, because manual tumour delineation is too time-consuming to perform daily. This project addresses that gap with an automated segmentation model based on the nnU-Net framework, trained on the UPENN-GBM dataset (630 de novo GBM patients) using T2 and FLAIR sequences, and evaluated against expert manual delineations on MR-Linac patient data. Results across two patients show Dice similarity coefficients of 0.84–0.89 maintained across fractions, with mean surface distances of 1.6–4.5 mm. A key finding is tumour migration of up to 19 mm between fractions — a shift large enough to compromise dose coverage — which automated tracking makes routinely detectable where manual delineation cannot. The longer-term aim is a clinical alert system that flags volumetric or positional changes exceeding predefined thresholds, supporting the clinical team in deciding whether plan adaptation is warranted.

Investigation of the effects of ultra-high dose rate "FLASH" on dosimeters' performance PhD active

Lead: Supaporn Srisuwan

Supervisor: Moshi Geso

The main goal of radiotherapy is to increase tumour control while minimizing normal tissue complications for the patient. Many advanced techniques have been developed to achieve this goal. A novel approach, ultra-high dose rate radiotherapy (FLASH-RT), delivers radiation at dose rates greater than 40 Gy/s within sub-seconds. This technique has been shown to reduce normal tissue toxicity while maintaining a tumour response comparable to that of conventional dose rate radiotherapy (approximately 0.03 Gy/s). However, the delivery of such high doses in sub-seconds poses significant dosimetric challenges. Conventional detectors suffer from saturation and ion recombination, leading to substantial errors and uncertainties in dose measurement. My research aims to investigate dosimeters capable of accurately determining radiation doses delivered by ultra-high dose rate beams. One approach involves employing photodisintegration gamma peaks generated by high-energy electrons proposed for FLASH radiotherapy and correlating them with the delivered dose. Another approach is the use of alanine dosimeters. Our results demonstrate that alanine dosimeters exhibit dose-rate independence when irradiated with ultra-high dose rate synchrotron X-ray radiation. In addition, ongoing studies are assessing the response of alanine dosimeters under ultra-high dose rate electron beams.

Liver tumour segmentation from MR-Linac active

Lead: Maryam Fallahpoor

This project focuses on liver tumor segmentation from magnetic resonance imaging (MRI), with a particular emphasis on the application of MR-Linac imaging data for clinically relevant treatment planning. The aim of this work is to develop robust and accurate deep learning–based segmentation approaches capable of delineating liver tumors in MR-guided radiotherapy settings, where challenges such as low soft-tissue contrast variability, motion artifacts, and anatomical deformation are common. By leveraging MR-Linac data, the research seeks to support adaptive radiotherapy workflows and improve the precision of radiation treatment delivery through reliable tumor localization and contouring in real-time or near real-time clinical environments.

Targeting low ADC regions for radiotherapy treatment of glioblastoma on an MR-Linac PhD active

Lead: Paowarin Khayaiwong

Supervisors: Ricky O'Brien , Pradip Deb , Daryl Wilding-McBride

Glioblastoma (GBM) is an aggressive brain tumour with a poor prognosis and low overall survival despite advances in medical treatment. As GBM remains incurable, current treatments may modestly prolong survival; however, in-field tumour recurrence during or after radiotherapy remains a major clinical challenge. Monitoring tumour progression during treatment is therefore essential. The emergence of Magnetic Resonance Imaging-Linear Accelerator (MRI-Linac) systems, which integrate diagnostic MRI with a linear accelerator, enables daily imaging and adaptive radiotherapy. The apparent diffusion coefficient (ADC), derived from MRI images acquired on MRI-Linac systems, is a crucial imaging biomarker for predicting tumour response. Changes in ADC during radiotherapy have been associated with tumour response and recurrence in GBM. This study focuses on low-ADC regions, as temporal ADC changes during treatment may correlate with tumour response. Low-ADC regions may therefore serve as suitable targets for dose-escalated adaptive radiotherapy using MRI-Linac systems.

Reconstructing dynamic 3D anatomy from sparse X-ray projections active

Lead: Xin Zhang

This project focuses on reconstructing dynamic 3D anatomy from sparse X-ray projections for applications in image-guided radiotherapy, CBCT, and orthopaedic imaging. Sparse radiographic views provide only incomplete observations of underlying anatomy, particularly when structures are moving, deforming, or articulated. To address this, I develop physics-aware learning frameworks that integrate continuous representations, anatomical priors, and differentiable X-ray rendering to recover patient-specific anatomy and motion directly from limited projection data.

Real-time tracking system for cardiac and respiratory motion during non-invasive stereotactic arrhythmia radioablation active

Lead: Wenjuan (Wendy) Xiong

Ventricular tachycardia is a life-threatening arrhythmia that is often challenging to treat safely. This project investigates a real-time tracking system for cardiac and respiratory motion during non-invasive stereotactic arrhythmia radioablation. The proposed technology aims to improve treatment precision, enabling more accurate radiation delivery, reducing complications, and ultimately improving patient outcomes.

A World Model for Adaptive Radiotherapy of Glioblastoma on the MR-Linac active

Lead: Daryl Wilding-McBride

Glioblastoma's poor prognosis reflects in part the limitations of fixed treatment plans applied to a tumour that evolves throughout the radiotherapy course. The MR-Linac makes visible what was previously hidden — fraction-by-fraction changes in tumour volume, morphology, and position — but current planning tools remain reactive, working only from today's anatomy with no capacity to anticipate how the tumour will respond to alternative treatment decisions. This project develops a world model — a learned simulator trained on longitudinal GBM MRI data — that takes the current tumour state observed on the MR-Linac and a proposed treatment action (dose, fractionation) as input, and predicts tumour appearance at the next fraction without delivering the treatment. This enables counterfactual reasoning over competing adaptive plans and, eventually, training of a reinforcement learning agent to discover optimal fraction-by-fraction strategies. The model is trained on the LUMIERE dataset (91 GBM patients, 638 longitudinal studies), with tumour state represented by volume, segmentation mask, diffusion signal, and MGMT methylation status. A first prototype demonstrates plausible prediction of the contrast-enhancing tumour region at the next timepoint. Key open questions include data volume requirements for generalisation, validation standards for clinical utility, and how to formalise treatment outcome as an optimisation objective.