New Brain Scan Could Help in Autism Detection

By DrOONeil (Own work) [CC BY-SA 3.0 (], via Wikimedia Commons

Many physicians and medical researchers believe they could improve the diagnosis and understanding of autism spectrum disorders if they had a reliable method of identifying specific abnormalities within the brain.

“Biomarkers” have proven to be elusive, often because the methods that were one promising with one group of patients does not work when it is applied to another. In a recent study, published in Nature Communications, scientists have experienced a surprising degree of success. Their proposed biomarker worked with a comparably high level of accuracy in evaluating two diverse sets of adults.

The Study

The technology discovered by the medical team was developed at the Advanced Telecommunications Research Institute International in Kyoto, Japan. The three major suppliers to the technology were from Brown University. The discovery is a computer algorithm called a “classifier” because it can classify sets of subjects, those with autism and those without.

The results of the classifier technology is based upon functional magnetic resonance imaging scans. By reviewing thousands of connections of brain networks in scores of individuals both with and without autism, the software detected 16 key interregional functional connections that allowed it to tell, with a high precision level, who had been diagnosed with autism and who hadn’t.

The team developed the classifier with 181 adult volunteer test subjects at three locations in Japan, and then applied it to a group of 88 American adults at seven locations. All of the volunteers in the study with an autism diagnosis had no intellectual disabilities.

Dr. Yuka Sasaki, research professor at Brown University states, “It is the first study to successfully apply a classifier to a totally different cohort. There have been numerous attempts before. We finally overcame the problem.”

The classifier blends two machine-learning algorithms and it worked well with each group, averaging around 85 percent accuracy with the Japanese volunteers and 75 percent with the Americans. The team calculated that the probability of seeing this degree of cross-population performance purely by accident was approximately 1.4 in a million. Dr. Mituso Kawato one of the co-authors of the study states, “These results indicate that although we developed a highly reliable classifier using the training data only in Japan, it is sufficiently universal to classify (autism) in the U.S.A. validation cohort.”

Can it Be Useful in a Clinical Setting?

The MRI scans required to gather the information were simple, Sasaki said. The test subjects only needed to spend about 10 minutes in the machine and did not have to undertake any elaborate activities or tasks. The person only had to lay still and rest.
Despite the data gathering being simple and even though the classified performed impressively well, as a matter of researcher, Sasaki said, it is not ready for use as a clinical tool. While the future might bring about that development, refinements to the technology will need to be made first.

ocd self test
Do you or a loved one feel like you might have a problem with OCD? Take the Self Test now to get more information.

The information provided on is designed to support, not replace, the relationship that exists between a patient/site visitor and his/her health professional. This information is solely for informational and educational purposes. The publication of this information does not constitute the practice of medicine, and this information does not replace the advice of your physician or other health care provider. Neither the owners or employees of nor the author(s) of site content take responsibility for any possible consequences from any treatment, procedure, exercise, dietary modification, action or application of medication which results from reading this site. Always speak with your primary health care provider before engaging in any form of self treatment. Click here to read our complete Terms of Use.

Susbscribe to our free newsletter for information & inspiration

Email Social