Breakthrough is going to help treatment and drug trials, scientists say.

Charles Marshall thinks he can now predict who will get dementia up to nine years before their symptoms appear, with a rate of accuracy of 82%.

But when Marshall and I talked, it turned out that nine-year figure actually understates the achievement. It simply represents the real-world time limit on his research project (so far). His work may predict dementia even further into the future.

He admits he and his team won't know until the cohort they have been studying lives longer, and more of them get dementia.

Marshall is a senior professor of neurology at the Queen Mary University of London, as well as a consultant in neurology at a London hospital.

He and his team have just published their findings in the peer-reviewed scientific journal Nature Mental Health. Their discovery may mark an important step forward on the (long) journey towards effective treatment, and a cure, of a disease that is killing millions and is always fatal.

Marshall and colleagues Sam Ereira, Sheena Waters and Adeel Razi found they could predict who would get dementia in the years ahead, and who wouldn't, by looking at certain MRI brain scans, known as functional MRIs or fMRIs.

Marshall says they looked especially at blood flows between different parts of the brain while the subject in question wasn't really doing anything - except lying in an MRI tube, daydreaming.

Neurologists refer to this state of mind as the default mode network or DMN, and it may be familiar to anyone who has studied "mindfulness," and become aware of how often we drift off into unfocused internal thought when there is nothing to distract us.

"Research shows that the DMN is implicated in several high-level cognitive processes such as social cognition and mental time-travel, resulting in a contemporary view that the DMN furnishes an individual with their narrative sense of self," Marshall and his team report.

The study was based on fMRIs and medical histories of over 1,100 people participating in the long-running U.K.

Biobank medical research project in Britain. The average age of the subjects was 70 years old and among the 103 who, so far, have developed dementia the average length of time between the functional MRI scans and medical diagnosis of dementia was 3.7 years.

It's when our brains are in this state that the early predictors of dementia show up most clearly in the fMRI scans, the researchers found.

Their data show that this is the most accurate early predictor yet, exceeding other tests including those using MRI scans and those using spinal taps.

It may be available in your local hospital within a couple of years, he says. (There is, as always, more research to do.)

Scientists are slowly learning more about dementia, a deadly disease that is a death sentence and which has no known cure and very few treatments of any kind. It is currently killing 6.7 million Americans and that figure is expected to double in the years ahead.

If doctors can diagnose patients with dementia early in the disease, before there are even noticeable symptoms, in the future that may help them treat it. At the moment there are a few drugs that can help. More will come. The earlier you catch a disease, the better.

Read: Credit scores and mortgage payments can help catch dementia early. Here's how.

Marshall adds that if we can accurately predict who is going to get dementia in advance, that also means there will be many more people who'd be willing to volunteer for drug trials.

There are other implications too, of course. If you were going to get dementia, would you want to know in advance? It will surely help more people put their affairs in order before it's too late, and make sure they can live their best life while they still can. It may also drum up business for Dignitas.

Having the option of knowing early must surely be better than not having it.