Magnetic resonance (MR) imaging has been used successfully to find and quantify gross functional deficits and neuroanatomical changes related to a number of neuropsychiatric disorders. However, often these results have been found using pooled populations of subjects within the test groups and within grossly-defined regions of the brain. One area where MR brain imaging has recently begun to show promise is in the autism spectrum of disorders (ASD). These are developmental disorders that probably involve a host of brain regions in their pathobiology. Functional MR (fMRI) studies indicate that there may be differences in regional brain function in response to several different face recognition and social attribute discrimination tasks. Recent work in our laboratory and elsewhere, focused on functionally- indicated zones such as the amgydala and the fusiform gyrus, have now begun to find group-wise structural differences between normal controls and ASD subjects. While these initial indications are promising, it remains the case, as noted above, that studies have been limited in terms of their ability to localize information spatially and to delineate subgroups within ASD. Our proposed efforts are centered on developing unique mathematical approaches that consider functional and structural information together in order to develop more sensitive and spatially-specific measurements. These approaches will: i.) estimate clustered regions of functional (fMRI-derived) parameters that adhere to particular gray matter structural zones, ii.) segment &measure key anatomical structure using object-neighbor and intensity constraints and iii.) nonrigidly register structural images from different subjects using both feature and intensity information. We will first perform confirmatory studies on normal controls. Finally, to show the effectiveness of our new approaches for studying a specific neuropsychiatric disorder, we will test their ability to separate subgroups in the autism spectrum (autism, Asperger's, PDD NOS) using both functional and structural measures.
Showing the most recent 10 out of 36 publications
