STELLA - Smart Technology to Extend Lives with Linear Accelerators
Radiotherapy is an essential pillar of cancer care for both curative and palliative cases; it is estimated that greater than 50% of all cancer cases globally can benefit from radiotherapy. Unfortunately, access to effective radiotherapy treatment is limited, especially in low- and middle-income countries (LMICs), where only 40% of those who require it have access to a radiotherapy treatment centre. Additionally, a variety of environmental and socio-economic factors contribute to significant downtime issues with current linear accelerators (LINACs) used for radiotherapy treatment, further exacerbating the gap in radiotherapy access between LMICs and high-income countries (HICs).
Project STELLA aims to combat this access gap by producing a cost-effective, robust radiotherapy LINAC for deployment in LMICs and elsewhere, providing accessible cancer treatment to LMICs without compromising on the quality or modernity. This also forms part of the International Atomic Energy Agency (IAEA) mandate on phasing out old Cobalt-60 machines, increasing international security over orphan sources.
The Oxford group, alongside Cambridge, forms Pillar 2 of STELLA. We are developing software for use in a future STELLA LINAC, including algorithms for fault-prediction in the LINAC and autosegmentation of tumours.
Multi-Leaf Collimator (MLC)
(Under Construction)
The multi-leaf collimator (or MLC) is a device used in radiotherapy LINACs to shape and modulate the intensity of a treating radiation beam. It consists of a number of individual "leaves", usually thin tungsten sheets, that move independently in and out of the beam path to allow for precise conformation of the radiation to the shape of the tumour, while limiting the dose to surrounding healthy tissue. MLCs are a key part of modern cancer treatment techniques, such as conformal radiotherapy and IMRT (Intensity-modulated radiation therapy).
MLCs are also the most fragile part of a modern radiotherapy LINAC and the cause of a significant proportion of LINAC downtime, especially in LMICs. This is due to the large number of small moving parts that are highly sensitive to environmental fluctuations and wear out over time.
As part of project STELLA, we aim to increase MLC robustness by analysing patterns of wear and failure modes in existing MLCs.
Fault Prediction
(Under Construction)
With new developments in machine learning for analysis and prediction of time-series data, STELLA aims to use radiotherapy machine log files to predict LINAC faults and failures prior to their occurrence. This has the potential to significantly reduce machine downtime, allowing proactive maintenance and repair of the LINAC.
Project Collaborators
- Prof Manjit Dosanjh
- Sam Leadley
- Prof Rajesh Jena (Cambridge)
- Dr Xin Du (Cambridge)
Past Students
- Alex Christie (Oxford, 2019-23)
- David Pugh (Oxford, 2020-24)