Autism is characterized by a broad spectrum of clinical manifestations including qualitative impairments in social interactions and communication, and repetitive and stereotyped patterns of behavior. Abnormal acceleration of brain growth in early childhood, signs of slower growth of neurons, and minicolumn developmental abnormalities suggest multiregional alterations. The aim of this study was to detect the patterns of focal qualitative developmental defects and to identify brain regions that are prone to developmental alterations in autism. Formalin-fixed brain hemispheres of 13 autistic (4–60 years of age) and 14 age-matched control subjects were embedded in celloidin and cut into 200-μm-thick coronal sections, which were stained with cresyl violet and used for neuropathological evaluation. Thickening of the subependymal cell layer in two brains and subependymal nodular dysplasia in one brain is indicative of active neurogenesis in two autistic children. Subcortical, periventricular, hippocampal and cerebellar heterotopias detected in the brains of four autistic subjects (31%) reflect abnormal neuronal migration. Multifocal cerebral dysplasia resulted in local distortion of the cytoarchitecture of the neocortex in four brains (31%), of the entorhinal cortex in two brains (15%), of the cornu Ammonis in four brains and of the dentate gyrus in two brains. Cerebellar flocculonodular dysplasia detected in six subjects (46%), focal dysplasia in the vermis in one case, and hypoplasia in one subject indicate local failure of cerebellar development in 62% of autistic subjects. Detection of flocculonodular dysplasia in only one control subject and of a broad spectrum of focal qualitative neuropathological developmental changes in 12 of 13 examined brains of autistic subjects (92%) reflects multiregional dysregulation of neurogenesis, neuronal migration and maturation in autism, which may contribute to the heterogeneity of the clinical phenotype.
Automatic classification is one of the basic tasks required in any pattern recognition and human computer interaction application. In this paper we discuss training probabilistic classifiers with labeled and unlabeled data. We provide a new analysis that shows under what conditions unlabeled data can be used in learning to improve classification performance. We also show that if the conditions are violated, using unlabeled data can be detrimental to classification performance. We discuss the implications of this analysis to a specific type of probabilistic classifiers, Bayesian networks, and propose a new structure learning algorithm that can utilize unlabeled data to improve classification. Finally, we show how the resulting algorithms are successfully employed in two applications related to human-computer interaction and pattern recognition; facial expression recognition and face detection.
Dimensionality reduction of a feature set is a common preprocessing step used for pattern recognition and classification applications. Principal Component Analysis (PCA) is one of the popular methods used, and can be shown to be optimal using different optimality criteria. However, it has the disadvantage that measurements from all the original features are used in the projection to the lower dimensional space. This paper proposes a novel method for dimensionality reduction of a feature set by choosing a subset of the original features that contains most of the essential information, using the same criteria as PCA. We call this method Principal Feature Analysis (PFA). The proposed method is successfully applied for choosing the principal features in face tracking and content-based image retrieval (CBIR) problems. Automated annotation of digital pictures has been a highly challenging problem for computer scientists since the invention of computers. The capability of annotating pictures by computers can lead to breakthroughs in a wide range of applications including Web image search, online picture-sharing communities, and scientific experiments. In our work, by advancing statistical modeling and optimization techniques, we can train computers about hundreds of semantic concepts using example pictures from each concept. The ALIPR (Automatic Linguistic Indexing of Pictures -Real Time) system of fully automatic and high speed annotation for online pictures has been constructed. Thousands of pictures from an Internet photo-sharing site, unrelated to the source of those pictures used in the training process, have been tested. The experimental results show that a single computer processor can suggest annotation terms in real-time and with good accuracy.
Objective
Younger siblings of children with autism spectrum disorder (ASD) are at high risk (HR) for developing ASD as well as features of the broader autism phenotype. While this complicates early diagnostic considerations in this cohort, it also provides an opportunity to examine patterns of behavior associated specifically with ASD compared to other developmental outcomes.
Method
We applied Classification and Regression Trees (CART) analysis to individual items of the Autism Diagnostic Observation Schedule (ADOS) in 719 HR siblings to identify behavioral features at 18 months predictive of diagnostic outcomes (ASD, atypical development, and typical development) at 36 months.
Results
Three distinct combinations of features at 18 months were predictive of ASD outcome: 1) poor eye contact combined with lack of communicative gestures and giving; 2) poor eye contact combined with a lack of imaginative play; and 3) lack of giving and presence of repetitive behaviors, but with intact eye contact. These 18-month behavioral profiles predicted ASD versus non-ASD status at 36 months with 82.7% accuracy in an initial test sample and 77.3% accuracy in a validation sample. Clinical features at age 3 among children with ASD varied as a function of their 18-month symptom profiles. Children with ASD who were misclassified at 18 months were higher functioning, and their autism symptoms increased between 18 and 36 months.
Conclusion
These findings suggest the presence of different developmental pathways to ASD in HR siblings. Understanding such pathways will provide clearer targets for neural and genetic research and identification of developmentally specific treatments for ASD.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.