New brain scan detects autism

Post Reply New Topic

A fifteen minute MRI brain scan, followed by computerised analysis, is said to identify small physical characteristics that detect autism with 90% accuracy (based on a sample of 20 people with and 20 people without autism). There is no clear indication of how long it would take to develop routine diagnosis for autism.

This kind of MRI procedure would cost between 250 and 450 dollars where I am, and has the benefit of objectivity.

Just 20 people with and without...?

Too small a group to makeany such claims. And any one study needs replication many times to be sure. Irresponsible science reporting as always. (And 90% even if true leaves 1 in 10 not detected.) Also wondering which subset of autistic people are and are not represented in the study (20 people is not nearly enough for a representative sample of all known causes and types of autism. Also wonder how many women represented. AS, autism, CDD, PDDNOS, essential, complex (and how many genetic syndromes represented), standard functioning labels... even these most official categories cannot be represented in 20 people let alone more accurate categories) Really bad science reporting. Yecch. And people will believe this too. Ew.

Obviously its a work in progress, and if it eventually could make diagnosis easier then it sounds like a positive.

Last edited by Free-Hinter-System on 10 Aug 2010, 5:43 pm, edited 1 time in total.

I want to get my brain scanned, now.

Me too! I would really like to see an image of my brain and what parts of it make me who I am. And I am strongly attracted to a completely objective, impartial diagnostic evaluation.

I had some detailed X-rays of my spine a few years ago, and the pictures had absolutely zero value in terms of treatment (because the treatment for back pain is pretty much the same, whatever the cause), but it was fantastic to see the precise cause of the pain - and of course (the real reason for the X-ray) the exclusion of serious disease. I would love to see a pain meter that accurately records the level of pain a person feels.

Since we still don't know exactly what causes Autism, or in which specific part(s) of the brain it resides, this is hocus-pocus that will only be used to deny services to anyone whose scan doesn't register a solid hit. I don't see a positive in it. There's a reason why the polygraph is inadmissible as evidence in court.

---

The above news report seems to me like another phony headline grabber. Within the last two years, I looked into the question as to whether a regular MRI in the USA would be able to accurately detect such challenges like ADHD, autism, and Asperger's and concluded, from talking with/corresponding with a number of doctors, that the best answer to the question was no. I have had multiple EEGs which are normal, a CT scan which is normal, and a regular MRI which is normal yet my ADHD Inattentive can be seen within two minutes or less by a skilled observer. The cost of the regular MRI to me (January 2009) at the HMO was about $1,200. Autism can be mild, moderate, or severe and I simply do not believe a regular MRI today can reliably and accurately detect autism in all customers at all. Again, another phony headline which misdescribes the state of 2010 brain scans (my view). I am very aware that a number of children and adults with Cerebral Palsy have normal brain scans. While brain scans are good, they are not perfect and there is no reason to fib/lie about what a regular MRI brain scan can actually see. That's my understanding.

Needs more research. But if it eventually is proven to be true diagnosis will be much simpler.

It could only have that effect if it was regarded with dogmatism, which obviously it would not, given that there is an acknowledged margin of error. Instead it would make things easier in many cases and not complicate them in the remainder.

Yes. Definitely a pilot study. Interesting, sure, but nowhere near "breakthrough" quality.

What worries me isn't the possibility of the 10% missed; it's the possibility of the false-positive rate. They scanned 20 people without autism; but that's not enough to tell what the false-positive rate is.

Let's say, for the sake of argument, that this is a very good test with only a 1% error rate. But do the math: That means that if you used it to screen everyone for autism, only around half the people who came up positive would actually be autistic! The other half would be NTs.

Also, I'm guessing they picked people who were obviously autistic for their study--people who fit the criteria very well--so as to be sure of picking autistic people. That kind of selection bias could mess up the results if they didn't take into account that there are more atypical autistics, borderline autistics, and autistics with comorbids that complicate their diagnosis, than there are "classic" cases. (Overwhelmingly more, in fact. My guess is probably that something like 90% of autistics are not "classic cases" of autism, either because they are PDD-NOS, because they are Asperger's, or because they have comorbid neurological issues.)

Additionally, we're not even really looking at this for telling apart NT and autistic. When someone goes in for an expensive thing like a brain scan, they have reason to believe something's up with them. So the test doesn't just have to distinguish between NT and autistic, which is hard enough; it has to distinguish between autistic and ADHD, autistic and mental retardation, autistic and brain injury, autistic and sensory integration disorder, autistic and schizophrenic, autistic and dyslexic... Getting the point yet? Telling apart autistic and NT is one thing; telling apart autistic and neurodiverse-in-general is quite another!

So yeah, celebrate that we've added another piece of information to the library; but remember that there are still a lot of shelves waiting to be filled.

I can understand the skepticism in this thread, but at least read the article before commenting.

This is a very serious study and potentially a very important breakthrough. I just saw the report on the BBC World News and it explained the findings better than the text, with better visuals. What the scan detects are minute differences in various areas of the brain that are found in people with autism as compared with people who are not autistic.

Any previous scan that anyone had is not relevant to this study because no one knew to look for these very tiny differences, which are not visible to the naked eye when looking at the scan.

The report also pointed out how difficult it is now to get an adult diagnosis of Asperger's, which is certainly true. One of the goals of this new technique is to make diagnosis much more available.

Other than the possibility that the study could turn out to have been too optimistic, I don't see any downside.

What Callista said. I wasn't saying the study is wrong. Just that the journalism is dreadful. Headlines like this have happened before. It never pans out the way headlines said it would. Willard is also correct to worry that if enough believed it it could act as a very destructive gatekeeper for services in a world that already tries hard to deny them for financial reasons.

It could also turn out to be like purported brain scans for schizophrenia (an even more nebulous category than autism). Often the purported brain changes in schizophrenia are actually caused by the most popular medications for it.

The two people interviewed in the report who had taken part in the study had Asperger's and had not been diagnosed until they were adults. They did not have "classic" low functioning autism. In fact, as they appeared in the report, talking about their experience growing up undiagnosed, they gave no indication of being in any way "odd".

I'm not saying I am 100% behind this study, but I would like to see responses to the actual report.

In the topic of diagnosis on Radiolab podcast, Jad Abumrad and Robert Krulwich discuss this method of diagnosing neurological differences. However, as they discussed it, the context was that of comparing the "diagnosed" individual to the "average" result, and determining change as a gauge of effectiveness for treatment paths. So I don't think it is as applicable to autism spectrum but as there isn't a treatment, if a concrete diagnostic method such as this is well developed it could be extremely useful to show the blatant neurological difference as a method of ruling out other explanations for atypical development & behavior.

New Scientist link MRI Autism article