Strategies for adaptive radiotherapy: towards clinically efficient workflows

Project completed in 2015.

Supervisors:
Dr Gareth Webster, University Hospitals Birmingham
Dr Kal Natarajan, University Hospitals Birmingham
Dr Hamid Dehghani, School of Computer Science
Professor Stuart Green, School of Physics

The current treatment paradigm for clinical radiotherapy is based on a treatment plan established using an imaging scan of the patient's anatomy prior to the start of treatment. Standardised treatment protocols, which define how much dose the tumour and healthy surrounding tissue can receive in order for the treatment to be safe and effective, are established based on the experience of clinicians and the interpretation of clinical trial results and historic data.

There are several problems with this situation: despite extensive efforts to restrict daily variations in patient shape and position, throughout the course of the several weeks that a radiotherapy schedule can last, substantial changes are often observed. Although positional shifts are now routinely corrected (image-guided radiotherapy, IGRT) and interventions made in cases of extreme changes (adaptive radiotherapy, ART), by far the most common scenario is for patient position to be assumed constant and for the radiation dose distribution received by the patient during treatment to be reported accordingly.

The impact of such assumptions can be clearly appreciated by considering the case of patients receiving radiotherapy for the treatment of prostate cancer: at University Hospital Birmingham, these patients are required to control the volumes of both their rectum and bladder (using daily micro-enemas and rigorous drinking protocols respectively) but significant positional errors of several centimetres are nevertheless seen. The impact of this is that the radiation dose received by the bladder and rectal walls, as well as to the small bowel, is associated with large uncertainties for any given patient. Since this situation is replicated over the entire population of such patients, attempts to develop robust predictive models for treatment outcome have so far been unsuccessful.

Although we know that applying today's standard treatment protocols in the context of these uncertainties leads to generally good outcomes across the patient population, we are also forced to accept that a sub-population of patients may receive a compromised treatment as the standard protocol is perhaps not the most appropriate for them. The current lack of individual predictive capability means that the next leap forward in modern radiotherapy - truly personalised treatment - cannot yet be safely realised.

The realisation of personalised radiotherapy treatment relies on two key developments: (i) robust multi-institutional data collection in terms of both detailed treatment outcomes and the radiation dose distribution received by the patient, (ii) for the quality of the data reported on the dose distribution to be substantially improved to take into account the anatomical changes discussed above. It is the second of these concerns that this project will investigate.

This project represents collaboration between PSIBS, Oncology and Radiotherapy Physics at University Hospital Birmingham and industry collaboration with Oncology Systems Limited (OSL, UK). OSL specialise in deformable image registration software, the application of which potentially allows daily variations in patient anatomy, as imaged at the time of treatment, to be co-registered so that the radiation dose received by the patient over the course of treatment can be accurately accumulated.

By the end of this PhD we hope to have contributed to the development of a validated clinical product for dose accumulation and to have demonstrated its clinical impact in terms of allowing more robust predictive models to be developed. In future, this work could facilitate both the ongoing monitoring of treatment delivery and the development of truly personalised radiotherapy treatments.

Link to ethesis: http://etheses.bham.ac.uk/6611/