AI, or artificial intelligence, is appearing more regularly in our daily lives as technology advances. Our home devices can now turn on our lights, wake us up, or start our coffee. While these small conveniences make our daily lives that much easier, there are many more complex ways to take advantage of this technology, especially within healthcare.
“CT is an incredibly important diagnostic tool, but it’s rarely used quantitatively,” said co-senior author Professor David Menon, from Cambridge’s Department of Medicine. “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.” Different types of blood in or around the brain can lead to different patient outcomes, and radiologists will often make estimates in order to determine the best course of treatment.
A deep learning tool based on an artificial neural network was developed by the investigators. On more than 600 separate CT scans, they conditioned the
instrument to display brain lesions of varying sizes and forms. It was possible for the AI to identify individual parts of each picture and to say whether or not it was natural.
This may be helpful for prospective research on the improvement of head trauma, as
the AI could be more reliable over time in identifying slight changes than a human.
Moreover the researchers say that in emergency rooms it might have a real use, helping to bring patients home faster.
Only between 10 and 15 percent of the patients who have a head injury have a lesion that can be seen on a CT scan. The AI could help classify those patients that require additional care, so that those without a brain lesion may be returned home, but it would be important to properly verify any clinical application of the instrument. The ability to automatically evaluate vast datasets would also enable researchers to address critical clinical research questions that have been difficult to address before, including the determination of specific prognostic features that could support target therapies in turn.