Researchers develop AI algorithm to detect & identify different brain injuries
Cambridge University researchers and colleagues from Imperial College London have developed an artificial intelligence (AI) algorithm that can detect and identify different types of brain injuries.
The AI has been clinically validated and tested on large sets of CT scans, and it was found that the algorithm was successfully able to detect, segment, quantify and differentiate different types of brain lesions.
The results, which were reported in The Lancet Digital Health, could be beneficial in large-scale research studies and for developing more personalised treatments for head injuries. It has also been suggested that with further validation, the artificial intelligence could be useful in certain clinical scenario - for instance, where radiological expertise is at a premium.
Once admitted to hospital, a patient who has suffered a head injury will usually be sent for a CT scan, which enables radiologists to check for blood in or around the brain and help to determine if surgery is required. Different types of blood in or around the brain lead to different patient outcomes, with radiologists often making estimates to determine the best type of treatment.
Co-senior author Professor David Menon, from Cambridge’s Department of Medicine, said: “CT is an incredibly important diagnostic tool but it’s rarely used quantitatively. Often, much of the rich information available in a CT scan is missed and as researchers we know that the type, volume and location of a lesion on the brain are important to patient outcomes.”
“Detailed assessment of a CT scan with annotations can take hours, especially in patients with more severe injuries,” added co-first author Dr Virginia Newcombe, also from Cambridge’s Department of Medicine.
“We wanted to design and develop a tool that could automatically identify and quantify the different types of brain lesions so that we could use it in research and explore its possible use in a hospital setting.”
How does the AI work?
The researchers have developed a machine learning tool based on an “artificial neural network”. The team trained the tool using over 600 different CT scans which showed brain lesions of various sizes and types, before validating the tool on an existing large dataset of CT scans.
The AI was able to classify individual parts of each image and distinguish whether it was normal or not. This could be particularly useful for future studies in how head injuries progress as it has been suggested the AI could be more consistent than a human when it comes to detecting subtle changes over time. Dr Newcombe said:
“This tool will allow us to answer research questions we couldn’t answer before. We want to use it on large datasets to understand how much imaging can tell us about the prognosis of patients.”
Professor Menon added: “We hope it will help us identify which lesions get larger and progress and understand why they progress so that we can develop more personalised treatment for patients in future.”
What are the benefits of using AI to detect different types of brain injury?
Currently the AI is to be used for research purposes only, but the researchers have said that with proper validation it could also be used in some clinical situations – for instance, if there are few radiologists in a resource-limited area.
The research team also believe it could potentially be used in emergency rooms, helping to discharge patients sooner. According to the study, of all the patients who have a head injury, only 10-15% have a lesion that can be seen on a CT scan. In this scenario, the AI could help identify these patients who need further treatment and those without a brain lesion can be sent home. However, the team has stressed that any future clinical use of the tool would need to be thoroughly validated.
Researchers will also be able solve vital clinical research questions that have not been answered previously - including the determination of relevant features for prognosis which in turn may help target therapies – through the ability to analyse such large data sets.
The research was supported in part by the European Union, the European Research Council, the Engineering and Physical Sciences Research Council, Academy of Medical Sciences/The Health Foundation and the National Institute for Health Research.