Personalising Image-Guidance for Paediatric Abdominal Radiotherapy with Artificial Intelligence

点击翻译按钮获取中文摘要

1) Brief description of the context of the research including potential impactThe target and tissue-sparing capabilities of proton beam therapy make it a promising modality to improve long-term outcomes in paediatric abdominal cancers, such as neuroblastoma, Wilms' tumour, and pelvic sarcomas. However, proton treatments are challenging to deliver because day-to-day variations in the contents of the gastrointestinal (GI) system, as well as weight fluctuations of the patient result in density changes that may lead to underdosage of the tumour and/or overshoot into healthy organs.In-room imaging allows to monitor day-to-day anatomical variation and may allow triggering the need for treatment adaption in-room. Extra volumetric imaging is often associated with additional radiation exposure that is particularly detrimental in children, due to the associated long-term increased risk of developing new cancers. Consequently, nowadays there is a lack of consensus between centres worldwide on what the most adequate image-guided protocolsfor children are, since it is of utmost importance to achieve a balance between information gained and exposure costs.This project aims to develop personalised image-guidance protocols in paediatric radiotherapy settings. We will combine artificial intelligence and different types of in-room imaging to monitor internal day-to-day anatomical change in bdominal proton beam therapy and propose novel workflows for image-guidance with fewer associated long-term risks.2) Aims and ObjectivesThe aim of this project is to develop novel image guidance protocols for paediatric protonbeam therapy that ensure radiation is delivered with high accuracy in the treatment ofabdominal cancers while reducing unnecessary diagnostic exposures to radiation.The specific objectives are to:- Investigate the optimal choice of imaging modalities available in room, combined withartificial intelligence image analysis methodology to detect internal, three-dimensionalanatomical change, through simulation studies.- Quantify the impact of anatomical change in the proton beam therapy dose delivered.- Propose, develop, and test a traffic light system for personalised image-guidedprotocols.- Evaluate the benefits of the methodology proposed in terms of imaging doses and theirassociated long-term risks.3) Novelty of Research MethodologyThe student will develop an artificial intelligence methodology applied to image-guidance workflows tailored to radiotherapy treatments in children. This is a novel methodological approach targeted to an under-researched clinical cohort.4) Alignment to EPSRC's strategies and research areasThis research is aligned with EPSRC's Theme of Healthcare Technologies and with the Research Areas of Medical Imaging and Artificial Intelligence Technologies.5) Any companies or collaborators involvedClinical paediatric oncologists from University College London Hospitals NHS Foundation Trust are collaborating on this research.