A diagnostic tool that yields an objective physiological assessment of whether someone is in pain, as opposed to relying on self-reported measures, is being developed by researchers at the Stanford University School of Medicine in Palo Alto, California. Using functional magnetic resonance imaging (fMRI) brain scans with advanced computer algorithms they accurately predicted thermal pain 81% of the time in healthy subjects, according to a study they reported in the 13 September issue of the online journal PLoS ONE.
However, they pointed out this was just a start, and more studies are now required to find out if their methods will work with different types of pain, such as chronic pain. Also, whether it is possible to distinguish, with an acceptable level of accuracy, between emotional states such as anxiety and depression and pain.
Pain is subjective in nature, and it is not unreasonable to assume a search for an objective measurement is nigh on impossible. However, the need for one is nearly universally acknowledged.
Senior investigator Dr Sean Mackey, associate professor of anesthesia and chief of the Division of Pain Management at the School, told the press:
“People have been looking for a pain detector for a very long time.”
“We rely on patient self-reporting for pain, and that remains the gold standard,” he added, explaining that he too, as a doctor treating patients with chronic pain, relies on their self-reporting.
But, he said, many patients ask, especially the very young and the very old, who find it hard to articulate their pain, wouldn’t it be great if there was a tool that could measure pain?
A June 2011 Institute of Medicine (IOM) report by a panel that included Mackey as a member estimates that more than 100 million Americans suffer chronic pain. This is associated with around 600 billion dollars a year in medical care costs and lost productivity.
They also found there is cultural bias against people who have chronic pain: they are seen as weak, and often perceived as lying about their pain. This complicates delivery of treatment, said the IOM panel.
Hank Greely, a Stanford law professor and expert on the legal, ethical and social issues surrounding the biosciences, said this bias also exists in the legal field, where hundreds of thousands of law suits a year hinge on the existence of pain.
“A robust, accurate way to determine whether someone is in pain or not would be a godsend for the legal system,” said Greely, who was not involved in the current study.
After attending a 2009 Stanford Law School event organized by Greely that brought together neuroscientists and legal scholars to debate how the neuroimaging of pain could be used and abused in the legal system, Mackey and two assistants from his lab decided to have a go and see if objective pain measurement was feasible.
Mackey said he was skeptical, but his two young lab assistans thought perhaps advances in neuroimaging methods meant there was a good chance they could come up with something. They said “we think we can do this. We would like to try,” said Mackey.
One of the assistants was co-author Neil Chatterjee, currently a MD/PhD student at Northwestern University. He said it was a bit of a whim, but they thought “maybe we can’t make the perfect tool, but has anyone ever really tried doing this on a very, very basic level?”
“It turned out to be surprisingly simple to do this,” said Chatterjee.
He and the other lab assistant, first author Dr Justin Brown, now an assistant professor of biology at Simpson College, came up with the idea in a discussion after the symposium.
For the first part of the study, 8 participants underwent brain-scanning while a heat probe was applied to their forearms, causing moderate pain.
The researchers recorded and intepreted, using advanced computer algorithms, scans of the brain patterns with and without pain. This enabled them to create a model of what pain looked like.
The computer model was based on an algorithm that was invented in 1995, called a a linear support vector machine (SVM). The researchers had the idea they could calibrate this using one set of participants, and then use it to accurately classify pain in another new set of participants.
So the second part of the study was another 16 participants underwent the same procedure as the first 8, but this time the researchers asked the “trained” computer to tell them, whether the new participants had thermal pain. It succeeded 81% of the time.
“… it did amazingly well,” said Chatterjee, “I was definitely surprised.”
He and his co-authors describe their experiment:
“Using eight individuals, we trained a linear SVM to distinguish these stimuli using whole-brain patterns of activity. We assessed the performance of this trained SVM model by testing it on 16 individuals whose data were not used for training. The whole-brain SVM was 81% accurate at distinguishing painful from non-painful stimuli (p