Close
Dr James Ruffle and the wider team at the High-Dimensional Neurology group in the Queen Square Institute of Neurology are transforming brain cancer care with AI-based tools.
AI technology plays an increasingly important role in our everyday lives. Whether it's planning routes with Google Maps or chatting with bots whilst shopping online, AI is making our lives easier in every way. And when it comes to our health, Artificial Intelligence has the potential to transform treatment and save thousands of lives.
Researchers at UCL are at the forefront of the AI healthcare revolution, working on projects which seek to make diagnosis, treatment and care easier for patients and clinicians.
One of those researchers is Dr James Ruffle (UCL Queen Square Institute of Neurology), who is developing AI technologies with the potential to revolutionise brain cancer care. This innovation is particularly significant for certain subtypes of brain cancer, where treatment outcomes have stagnated for nearly three decades.
Dr Ruffle said: “AI technology presents a critical opportunity in healthcare, where instead of merely treating the disease, it ushers in a new era where we can treat the patient in a more individual way. The opportunities presented by AI in personalised medicine are particularly promising and may lead to genuine enhancements in brain cancer care.”
“At present, there is very limited – if any – use of AI in the diagnosis and treatment of NHS brain cancer care. However, we expect this to change swiftly, as it is one of the most significant areas of research and development, having even been highlighted and prioritised by the UK Government.”
With 100,000 people living with brain cancer in the UK, there is no one size-fits-all approach to treatment. The unique position and size of every individual’s brain tumour means that every patient’s scan must be analysed closely by a radiologist. This analysis helps determine whether a patient’s tumour has been successfully removed following surgery, reduced in size after chemotherapy or radiotherapy, or whether, despite treatment, it has sadly grown. The complexity and individuality of each tumour’s appearance makes this no easy task and mistakes are possible.
“Sometimes, these appearances are unclear, making it difficult to be certain,” Dr Ruffle explained, “Since these decisions rely on visual review, interpretations can differ between clinicians reading the brain scan, even among world-leading neuroradiologists.”
As a solution to this problem, Dr Ruffle and the High-Dimensional Neurology research team have developed an AI-based tool called VASARI-auto. This tool can analyse brain scans within seconds and could be used to predict an individual patient’s treatment outcome more accurately than by a doctor’s analysis alone.
“Here, AI presents an opportunity to enhance the neuroradiologist’s interpretation by offering improved precision. For example, in our own research, we found such models could not only accurately delineate – or ‘segment’ – a brain tumour with millimetre precision, but it could typically do it within less than a second,” Dr Ruffle said.
As part of the research programme, the UCL Queen Square Institute of Neurology and University College London Hospital team tested the VASARI-auto tool on 1,172 patient scans, producing accurate results across patients of all ages and sexes.
Additionally, a workforce analysis forecast that if deployed nationally, VASARI-auto could save more than £1.5million in NHS costs within the next three years.
This is promising news for both the NHS and brain cancer patients, although we may still be some way off from seeing VASARI-auto used at hospitals around the country just yet.
“AI is not a simple toy that can be developed quickly, nor should its deployment in patient care be rushed,” Dr Ruffle said, “This is a primary reason why many of the highly promising tools under research and development take time to transition into clinical practice. With researchers and healthcare professionals bearing the responsibility to provide high-quality care, it is crucial that we exert every effort to ensure these models are sufficiently robust, resilient, and equitably powerful so that they can deliver clear benefits to all patients equally.”
As we wait for the deployment of this exciting new technology, there are plenty of other innovative AI-tools in development at UCL. These tools have the potential to revolutionise the entire patient pathway, bringing patient care into the future.
In the High-Dimensional Neurology lab, where Dr Ruffle works, a multidisciplinary team is developing advanced radiology tools that precisely outline a patient’s tumour. These tools aid both the initial planning of personalised targeted therapies and the subsequent monitoring scans a patient undergoes after treatment.
In the Neurosurgical Department, robotic-assisted surgery offers a route for new intra-operative decision-making tools that may enhance the chances of treatment success.
Meanwhile, in the Department of Neuropathology, AI technologies are being developed to analyse complex microscopy images and provide detailed tissue diagnoses.
“Although these initiatives are currently under rigorous development across UCL, the shared ambition among all groups is to see this work translated into the clinical frontline and to bring benefit to patients,” Dr Ruffle explained.
So how far are we away from this new reality?
“I believe the future is bright for AI in brain cancer treatment,” said Dr Ruffle, “In the next decade, I suspect we will begin to see more of these promising tools moving towards routine clinical use, whether in initial patient triage to the teams who need to treat them, when a patient undergoes medical imaging with radiology for their initial diagnosis and treatment planning, or perhaps for medical imaging when the healthcare team are monitoring a patient’s response to treatments such as chemotherapy or radiotherapy.
“It is quite likely that this advantage could even be leveraged in future research to develop new innovative treatments for brain cancer.”
Image credit: Research paper in NeuroImage: Clinical