Speech stimulation during functional MR imaging as a potential indicator of autism

Radiology. 2011 Aug;260(2):521-30. doi: 10.1148/radiol.11101576. Epub 2011 May 31.


Purpose: To determine the feasibility of applying functional magnetic resonance (MR) imaging as an objective indicator of language disability in autism by using passive speech stimulation.

Materials and methods: This prospective study was approved by the institutional review board, and informed consent was obtained from the parents or guardians of all subjects. Functional MR imaging was performed during passive presentations of prerecorded speech in 15 control subjects (mean age ± standard deviation, 12.1 years ± 4.3) and 12 language-impaired, age-matched autistic subjects (mean age, 12.4 years ± 4.7). An additional 27 autistic children (mean age, 8.4 years ± 3.1), who underwent imaging while sedated with propofol as part of routine clinical MR evaluations, were also included. Activation maps for each subject were computed by using univariate general linear model analyses. The spread (quantified as number of voxels) and amplitude of the functional MR imaging activation were then quantified within two anatomically specified regions of interest known to be involved with language: the primary auditory cortex (A1) and the superior temporal gyrus (STG). Group differences were compared by using analysis of variance, two-sample t tests, and Wilcoxon rank sum tests where appropriate. The threshold for autism was defined as 1 standard deviation below the control mean for subjects imaged in the alert state. A similar threshold was estimated for sedated autistic subjects on the basis of differences between nonsedated and sedated autistic subjects.

Results: Activity in A1 did not differ between autistic and control subjects. However, mean amplitude and spread of activity in the STG differed between autistic and control subjects (P < .001). Values for 10 of the 12 (83%) nonsedated autistic subjects decreased at least 1 standard deviation below the control distribution. The threshold derived from sedation-adjusted values of the control group enabled identification of 26 of the 27 (96%) sedated autistic subjects.

Conclusion: Functional MR imaging activation within the STG in response to passive speech stimulation helped differentiate autistic from control subjects, demonstrating the potential utility of functional MR imaging as an objective indicator of language impairment in autism. Future studies may lead to an early and objective indicator for autism with these methods.

MeSH terms

  • Adolescent
  • Analysis of Variance
  • Autistic Disorder / physiopathology*
  • Brain / physiopathology*
  • Case-Control Studies
  • Child
  • Child, Preschool
  • Female
  • Humans
  • Image Processing, Computer-Assisted
  • Linear Models
  • Magnetic Resonance Imaging / methods*
  • Male
  • Prospective Studies
  • Speech*
  • Statistics, Nonparametric