Treating cancer patients in Beijing with radiotherapy powered by machine learning

Dr Adam Szmul (UCL Faculty of Engineering Sciences) used UCL-Peking University Strategic Partner Funds to take adapted radiotherapy methods to a cancer hospital in China.

Radiotherapy is used to treat approximately 50% of people who are diagnosed with cancer. It works by delivering high energy beams to a patient’s body to destroy cancer cells. The treatment requires imaging to identify which regions of the body should receive the treatment, as well as ‘healthy’ parts of the body where radiation should be avoided. A patient's radiotherapy treatment plan is usually prepared using a Computed Tomography (CT) scan. 

Radiotherapy treatment is delivered over a period of few weeks, during which a patient’s body might change due to treatment and physiological processes. Radiotherapy machines are equipped with on-board Cone Beam Computed Tomography (CBCT) imaging devices, which provide images of inferior quality to CT scan quality. CBCTs can capture images of the patient at the time of treatment to correctly position the individual during the treatment, and to monitor if the original treatment plan is still valid. In current clinical practice, if significant changes are observed in the patient's anatomy – rendering the original treatment plan unsafe for delivery – the patient must undergo rescanning with a CT. This is followed by the preparation of a new treatment plan. However, a new plan could potentially be prepared using CBCTs, if their quality was sufficient for treatment planning.  

Dr Adam Szmul is part of a team at UCL which has developed a machine learning method for improving the quality of CBCT images. This method puts CBCT image quality on a par with the quality of CT images, which means radiotherapy dosing can also be accurately calculated. Wishing to demonstrate how this method could be used to help more patients across the world, Adam successfully applied for UCL-Peking University Strategic Partner Funds. This enabled him to collaborate with researchers at Peking University, who could then apply the method at Peking University Cancer Hospital.

Rolling out more targeted radiotherapy treatment

“Radiotherapy is a balance of giving the treatment to the locations where cancer is, but sparing the organs which shouldn't receive radiation because it could cause damage,” Adam explained. “Another factor to consider is that a patient’s body can change over time. Radiotherapy is usually delivered over a few weeks. There might be changes due to the treatment, like a patient might lose weight. There are also changes to the body which happen naturally, like bowel movements and the bladder filling.”

As adequate technology did not previously exist to get accurate imagery of a patient’s body at the time of treatment, Adam and his colleagues used machine learning to develop the tool for adaptive radiotherapy (ART). This enables medical teams to better see how a patient’s body looks on the day of the treatment, and adjust the treatment plan as necessary. “This is all about delivering the best possible treatment for patients, for their anatomy, on each treatment day,” Adam said.

As the team had already used the method they developed for patient data at UCLH, they wanted to work in other clinical settings around the world to see if the approach could be transported elsewhere. Peking University Cancer Hospital has an experienced team of clinical collaborators, with plentiful facilities and a lot of clinical data. Adam thought they would be a good match for this project. He was due to visit Beijing to take this forward through UCL-Peking University Strategic Partner Funds. However, due to ongoing travel restrictions from the UK to China after Covid-19, UCL hosted a researcher, Mr Zhengkun Dong, from Professor Yibao Zhang’s group at Peking University, instead. During this time, Adam was able to share his knowledge about ART, for the researcher to take back to implement at Peking University Cancer Hospital.

Developing globally relevant methods

According to Adam, the project was a great success. “We’re still awaiting the clinical outcomes of the work, but our methods have been successfully applied at Peking University Cancer Hospital,” he said. “It shows that we’re developing methods at UCL that can be used anywhere.”

The team has had one paper published about their work, and they are currently finalising a manuscript for another. The collaborators are still in regular contact to discuss how they can push their work forward, and this project enabled them to seek funding from other sources, with one grant proposal currently under revision. It has also facilitated new collaborations between other researchers from UCL, which has very recently been awarded a grant by National Natural Science Foundation of China (NSFC) and The Royal Society in the UK.

“We wanted to prove that the work we’re doing is not just an academic thing,” Adam said. “We wanted to show that it’s clinically applicable, and can be used to help more people. And we’ve definitely done that. For me, it was a great experience professionally and personally. Managing the funding and the project will be good for my professional development. It was also really enjoyable hosting someone from another country, learning about their culture, plus showing them our way of living and doing research. I very much appreciate the opportunity.”

Links

• Simulation of a new respiratory phase sorting method for 4D-imaging using optical surface information towards precision radiotherapy
• Dr Adam Szmul's profile
• UCL Faculty of Engineering Sciences
• Find out more about funding opportunities offered by UCL Global Engagement

Featured image

• Patient receiving medical treatment for head & neck cancer
• Credit: iStock