A computer does better than a doctor at diagnosing certain brain diseases, research has suggested.
Blue areas show volume loss picked up by computer
Experts taught a standard computer how to diagnose Alzheimer's from brain scans, and got a 96% success rate.
The accuracy of diagnosis from standard scans, blood tests and interviews carried out by a clinician is 85%.
The findings, published in the journal Brain, could lead to earlier diagnosis and more successful treatment of dementias, say scientists.
Researchers from the Wellcome Trust Centre for Neuroimaging at University College London say computers have several advantages for diagnosing Alzheimer's - a condition caused by the build-up of plaques and tangles of tissue in the brain.
Professor Richard Frackowiak said the computers were better able to distinguish signs of Alzheimer's than humans, and proved cheaper, faster and more accurate than current methods.
"It's beginning to look like it will have to come into clinical practice," he said. "Machines are clearly able to do that sort of thing better."
The method involves teaching a standard computer the difference between brain scans from patients with proven Alzheimer's disease and people with no signs of the disease at all.
The two conditions can be distinguished with a high degree of accuracy on a single clinical MRI scan without the need for time consuming follow-up tests, say the scientists.
They think the technique will be particularly useful for reassuring elderly people with mild memory loss that they are not suffering from Alzheimer's disease.
It may also allow researchers to study progression of the disease in a patient, and perhaps eventually lead to a way of screening new drugs.
"In the long-run, we'd like to use these techniques as ways of classifying patients with something like a degenerative disease into various stages," explained Professor Frackowiak.
"From the point-of-view of developing new pharmaceuticals for these disorders there's great potential," he added.