Consultant radiologist Dr Christina Messiou
Although the concept of artificial intelligence (AI) has now existed since 1956, it's only now that its use has become the norm. Earlier this year, Prime Minister Theresa May put AI and big data in the spotlight when she called for technology to transform how the UK prevents, diagnoses and treats cancer.
Big data is often cited as the fule that powers AI, and healthcare providers are uniquely placed to use the large amounts of data collected from patients as part of their care, as well as information gathered from clinical trials. At The Royal Marsden, our experts are exploring the lastest advances in technology to help diagnose cancer earlier and improve the quality of life for patients living with cancer.
As part of a groundbreaking project, Dr Christina Messiou, Consultant Radiologist, is working with Professor Andrea Rockall and Dr Ben Glocker from Imperial College London to develop machine learning that will assist radiologists in reporting whole-body MRI scans in patients with Myeloma. Machine learning is a type of AI in which computers are taught how to do things independently - for exmaple, to identify scans that show evidence of cancer from healthy images.
"If we want to truly understand a patient's cancer, we often need to assimilate vast amounts of information, drawing on huge and complex datasets" says Dr Messiou. "Radiology has always been a technology-driven specialism and AI is another really powerful tool to help us do our jobs better"
With colleagues at Imperial College London, we're developing state-of-the-art technology that will enable us to analyse this information much faster
Improving quality of life
Unlike CT scans, whole-body MRI scans can detect cancer in the bone marrow before it has caused destruction of the outer bone cortex, meaning a diagnosis can be made much earlier. This is particularly important in myeloma because, as the disease progresses, it can result in irreparable bone damage, leading to debilitating complications.
"The challenge with using whole-body MRI scans is that they generate a vast amoung of data - around 1,000 images for each scan, which need to be examined individually by a radiologist," says Dr Messiou. "This is where we're turning to AI, to remove some of these more time-consuming tasks. By developing a system that can flag up sites of possible disease and measure the amount of disease, we can get to work quicker on diagnosing and treating our patients, guided by extremely precise measurements."
The new study with Imperial, MALIMAR (Machine Learning in Myeloma Response), funded by the National Institute for Health Research (NIHR), will train a computer algorithm to recognise the difference between a healthy scan and a patient with myeloma. Researchers will then examine the time it takes to process and report whole-body MRI scans normally, compared with a radiologist using machine-learning technology. The study aims to find a way to quatify the amount of diseases visible on whole-body MRIs.
"Big data is collaborative by its very nature, and this study will draw on expertise from radiologists, physicists, data scientists and computer scientists, as well as the machine-learning group at Imperial," says Professor Rockall, who is an Honorary Consultant Radiologist at The Royal Marsden and Clinical Chair in Radiology at Imperial.
We hope AI will not only detect and quantify the amount of disease, but also reduce the time needed for a radiologist to review the entire scan. This might allow whole-body MRI to become a test that can be offered to myeloma patients throughout the NHS
Our researchers are also developing big-data computing infrastructure that will underpin all research at the Trust, across all cancer types. Professor Winette can der Graaf, who is leading this work as part of the Digital theme at the NIHR Biomedical Research Centre at The Royal Marsden Foundation Trust and The Institute of Cancer Research, says: "Our vision is to provide smarter, kinder treatments, tailored to individual patients. This can only be achieved through rapidly accessing all relevant data for a patient and utilising smart analysis tools to support clinical decision-making."
A new data management system called PROFILES, which has been used in the Netherlands since 2009, will collect anonymous data on Patient Reported Outcome Measures (PROMs). "The challenge is embedding this technology and capturing this information in a routine way, as part of a patient's clinical records, in order to guide our clinical decisions," says Professor van der Graaf. "In the shorter term, we're already using this technology to better understand the care needs of young adults with cancer."
Professor van der Graaf, Dr Olga Husson and colleagues have written to all patients ages 25-39 who have been diagnosed with cancer in the past five years at The Royal Marsden and other hospitals in London, Ipswich and Southampton. Through focus groups, interviews and surveys, the study will measure care needs and experiences, quality of life, psychological distress and wider impact of cancer.
"These patients are confronted with a life-threatening disease, alongside the usual challenges or careers, finances, partners, fertility and young children," says Professor van der Graaf. "By interrogating large amounts of data, we want to move oncology into the digital age by exploring young people's digital preferences, and develop clinical pathways of the future."
The results will be used to design a tool or service, such as a digital app, to support young adults with cancer at the Trust.
The possibilities for harnessing this technology in cancer care are endless, and The Royal Marsden is just beginning to explore them.