EEGs May Diagnose Autism Spectrum Disorders in Infants

A standard electroencephalogram (EEG) and machine-learning algorithms have been combined to form an experimental noninvasive test to evaluate an infant's autism risk.

Researchers at Children's Hospital Boston CHB; MA, USA) recorded resting EEG signals from 79 babies aged 6-24 months of age participating in a larger study aimed at finding very early risk markers of autism; 46 of the infants had an older sibling with a confirmed diagnosis of an autism spectrum disorder (ASD); the other 33 had no family history of ASD. As the babies watched a research assistant blowing bubbles, EEG recordings were made via a 64-electrode cap. When possible, the tests were repeated at 6, 9, 12, 18, and 24 months of age.

The researchers took the EEG brain-wave readings for each electrode and computed their modified multiscale entropy (mMSE)--a measure borrowed from chaos theory that quantifies the degree of randomness in a signal, from which characteristics of whatever is producing the signal can be inferred. The investigators looked at the entropy of each EEG channel, which is believed to contain information about the density of neural connections in the brain region near that electrode, how connections between them are organized, and the balance of short- and long-distance connections. By following the high-risk group over time, and comparing EEG patterns in those who receive an actual ASD diagnosis and those who appear to be developing normally, the researchers hope to identify specific ASD patterns.

The researchers have already found that at 9 months, babies undergo important changes in their brain function that are critical for the emergence of higher-level social and communication skills, skills often impaired in ASDs. For unclear reasons, they found a gender difference; classification accuracy was greatest for girls at 6 months and remained high for boys at 12 and 18 months. Overall, however, the distinction between the high-risk group and controls was smaller when infants were tested at 12 to 24 months. The authors speculate that the high-risk group may have a genetic vulnerability to autism that can be influenced and sometimes mitigated by environmental factors. The study was published on February 22, 2011, in the online open-access journal BMC Medicine, a publication of BioMedCentral.

"Electrical activity produced by the brain has a lot more information than we realized; computer algorithms can pick out patterns in those squiggly lines that the eye can't see,” said lead author William Bosl, PhD, a neuroinformatics researcher in the CHB Informatics Program. "With enough data, I'd like to follow each child's whole trajectory from 6 to 24 months. The trend over time may be more important than a value at any particular age.”

"Many neuroscientists believe that autism reflects a 'disconnection syndrome,' by which distributed populations of neurons fail to communicate efficiently with one another,” added coauthor Charles Nelson, PhD, research director of the developmental medicine center at CHB. "The current paper supports this hypothesis by suggesting that the brain of infants at high risk for developing autism exhibit different patterns of neural connectivity, though the relationship between entropy and the density of neural arbors remains to be explored.”

Related Links:
Children's Hospital Boston
BioMedCentral