Will EEG Data Analysis Help Diagnose Autism?

by

Com­put­er analy­sis of EEG pat­terns sug­gests a poten­tial diag­nos­tic test for autism (Eure­ka Sci­ence News):

“Wide­ly avail­able EEG test­ing can dis­tin­guish chil­dren with autism from neu­rotyp­i­cal chil­dren as ear­ly as age 2, finds a study from Boston Chil­dren’s Hos­pi­tal. The study is the largest, most rig­or­ous study to date to inves­ti­gate EEGs as a poten­tial diag­nos­tic tool for autism, and offers hope for an ear­li­er, more defin­i­tive test. Researchers Frank H. Duffy, MD, of the Depart­ment of Neu­rol­o­gy, and Hei­delise Als, PhD, of the Depart­ment of Psy­chi­a­try at Boston Chil­dren’s Hos­pi­tal, com­pared raw EEG data from 430 chil­dren with autism and 554 con­trol sub­jects, ages 2 to 12, and found that those with autism had con­sis­tent EEG pat­terns indi­cat­ing altered con­nec­tiv­i­ty between brain regions — gen­er­al­ly, reduced con­nec­tiv­i­ty as com­pared with controls.”

Study: A sta­ble pat­tern of EEG spec­tral coher­ence dis­tin­guish­es chil­dren with autism from neu­ro-typ­i­cal con­trols — a large case con­trol study (BMC Med­i­cine). Pro­vi­sion­al Abstract:

  • Back­ground: The autism rate has recent­ly increased to 1 in 100 chil­dren. Genet­ic stud­ies demon­strate poor­ly under­stood com­plex­i­ty. Envi­ron­men­tal fac­tors appar­ent­ly also play a role. Mag­net­ic res­o­nance imag­ing (MRI) stud­ies demon­strate increased brain sizes and altered con­nec­tiv­i­ty. EEG coher­ence stud­ies con­firm con­nec­tiv­i­ty changes. How­ev­er, genetic‑, MRI‑, and/or EEG-based diag­nos­tic tests are not yet avail­able. The var­ied study results like­ly reflect method­olog­i­cal and pop­u­la­tion dif­fer­ences, small sam­ples, and for EEG, lack of atten­tion to group-spe­cif­ic artifact.
  • Con­clu­sions: Clas­si­fi­ca­tion suc­cess sug­gests a sta­ble coher­ence load­ing pat­tern that dif­fer­en­ti­ates ASD- from C‑group sub­jects. This might con­sti­tute an EEG coher­ence-based phe­no­type of child­hood autism. The pre­dom­i­nant­ly reduced short-dis­tance- coher­ences may indi­cate poor local net­work func­tion. The increased long-dis­tance- coher­ences may rep­re­sent com­pen­sato­ry process­es or reduced neur­al prun­ing. The wide aver­age spec­tral range of fac­tor load­ings may sug­gest over-damped neur­al networks.

Relat­ed articles: