Feature selection and classification of imbalanced datasets Application to PET images of children with autistic spectrum disorders

Learning with discriminative methods is generally based on minimizing the misclassification of training samples, which may be unsuitable for imbalanced datasets where the recognition might be biased in favor of the most numerous class. This problem can be addressed with a generative approach, which typically requires more parameters to be determined leading to reduced performances in high dimension. In such situations, dimension reduction becomes a crucial issue. We propose a feature selection/classification algorithm based on generative methods in order to predict the clinical status of a highly imbalanced dataset made of PET scans of forty-five low-functioning children with autism spectrum disorders (ASD) and thirteen non-ASD low functioning children. ASDs are typically characterized by impaired social interaction, narrow interests, and repetitive behaviors, with a high variability in expression and severity. The numerous findings revealed by brain imaging studies suggest that ASD is associated with a complex and distributed pattern of abnormalities that makes the identification of a shared and common neuroimaging profile a difficult task. In this context, our goal is to identify the rest functional brain imaging abnormalities pattern associated with ASD and to validate its efficiency in individual classification. The proposed feature selection algorithm detected a characteristic pattern in the ASD group that included a hypoperfusion in the right Superior Temporal Sulcus (STS) and a hyperperfusion in the contralateral postcentral area. Our algorithm allowed for a significantly accurate (88%), sensitive (91%) and specific (77%) prediction of clinical category. For this imbalanced dataset, with only 13 control scans, the proposed generative algorithm outperformed other state-of-the-art discriminant methods. The high predictive power of the characteristic pattern, which has been automatically identified on whole brains without any priors, confirms previous findings concerning the role of STS in ASD. This work offers exciting possibilities for early autism detection and/or the evaluation of treatment response in individual patients.

[1]  Johan D. Carlin,et al.  The neural basis of eye gaze processing , 2013, Current Opinion in Neurobiology.

[2]  Masoud Nikravesh,et al.  Feature Extraction - Foundations and Applications , 2006, Feature Extraction.

[3]  Josef Kittler,et al.  Floating search methods in feature selection , 1994, Pattern Recognit. Lett..

[4]  Jean-Baptiste Poline,et al.  Inverse retinotopy: Inferring the visual content of images from brain activation patterns , 2006, NeuroImage.

[5]  J. Suckling,et al.  Mapping the brain in autism. A voxel-based MRI study of volumetric differences and intercorrelations in autism. , 2004, Brain : a journal of neurology.

[6]  G. Schwarz Estimating the Dimension of a Model , 1978 .

[7]  E. Courchesne,et al.  The brain response to personally familiar faces in autism: findings of fusiform activity and beyond. , 2004, Brain : a journal of neurology.

[8]  Kim M. Dalton,et al.  Gaze fixation and the neural circuitry of face processing in autism , 2005, Nature Neuroscience.

[9]  Steven C. R. Williams,et al.  Describing the Brain in Autism in Five Dimensions—Magnetic Resonance Imaging-Assisted Diagnosis of Autism Spectrum Disorder Using a Multiparameter Classification Approach , 2010, The Journal of Neuroscience.

[10]  B Mazoyer,et al.  Regional cerebral blood flow in childhood autism: a SPECT study. , 1992, The American journal of psychiatry.

[11]  E. Bullmore,et al.  The amygdala theory of autism , 2000, Neuroscience & Biobehavioral Reviews.

[12]  G. McCarthy,et al.  Neural basis of eye gaze processing deficits in autism. , 2005, Brain : a journal of neurology.

[13]  Nitesh V. Chawla,et al.  Editorial: special issue on learning from imbalanced data sets , 2004, SKDD.

[14]  Jean-Francois Mangin,et al.  Classification Based on Cortical Folding Patterns , 2007, IEEE Transactions on Medical Imaging.

[15]  N. Hadjikhani,et al.  Anatomical differences in the mirror neuron system and social cognition network in autism. , 2006, Cerebral cortex.

[16]  Janaina Mourão Miranda,et al.  Investigating the predictive value of whole-brain structural MR scans in autism: A pattern classification approach , 2010, NeuroImage.

[17]  J B Poline,et al.  Temporal lobe dysfunction in childhood autism: a PET study. Positron emission tomography. , 2000, The American journal of psychiatry.

[18]  E H Cook,et al.  Quantifying the phenotype in autism spectrum disorders. , 2001, American journal of medical genetics.

[19]  David R. Anderson,et al.  Bayesian Methods in Cosmology: Model selection and multi-model inference , 2009 .

[20]  Y.S. Hung,et al.  Gene selection for Brain Cancer Classification , 2006, 2006 International Conference of the IEEE Engineering in Medicine and Biology Society.

[21]  Y. Samson,et al.  Perception of complex sounds in autism: abnormal auditory cortical processing in children. , 2004, The American journal of psychiatry.

[22]  Christopher F. Chabris,et al.  Activation of the fusiform gyrus when individuals with autism spectrum disorder view faces , 2004, NeuroImage.

[23]  Leo Breiman,et al.  Random Forests , 2001, Machine Learning.

[24]  N. Graham,et al.  Areas beneath the relative operating characteristics (ROC) and relative operating levels (ROL) curves: Statistical significance and interpretation , 2002 .

[25]  Geoffrey J McLachlan,et al.  Selection bias in gene extraction on the basis of microarray gene-expression data , 2002, Proceedings of the National Academy of Sciences of the United States of America.

[26]  L. Lotspeich,et al.  White matter structure in autism: preliminary evidence from diffusion tensor imaging , 2004, Biological Psychiatry.

[27]  Isabelle Guyon,et al.  An Introduction to Variable and Feature Selection , 2003, J. Mach. Learn. Res..

[28]  Eric Courchesne,et al.  Outcome classification of preschool children with autism spectrum disorders using MRI brain measures. , 2004, Journal of the American Academy of Child and Adolescent Psychiatry.

[29]  N. Minshew,et al.  MRI volumes of amygdala and hippocampus in non–mentally retarded autistic adolescents and adults , 1999, Neurology.

[30]  P. Massart,et al.  Minimal Penalties for Gaussian Model Selection , 2007 .

[31]  Ron Kohavi,et al.  A Study of Cross-Validation and Bootstrap for Accuracy Estimation and Model Selection , 1995, IJCAI.

[32]  Federico Girosi,et al.  Support Vector Machines: Training and Applications , 1997 .

[33]  Ron Kohavi,et al.  Wrappers for Feature Subset Selection , 1997, Artif. Intell..

[34]  Ralph-Axel Müller,et al.  Abnormal activity patterns in premotor cortex during sequence learning in autistic patients , 2004, Biological Psychiatry.

[35]  R. Schultz Developmental deficits in social perception in autism: the role of the amygdala and fusiform face area , 2005, International Journal of Developmental Neuroscience.

[36]  Andrew L. Alexander,et al.  Diffusion tensor imaging of the corpus callosum in Autism , 2007, NeuroImage.

[37]  Ron Kohavi,et al.  Wrappers for feature selection , 1997 .

[38]  R. Tibshirani Regression Shrinkage and Selection via the Lasso , 1996 .

[39]  Chih-Jen Lin,et al.  LIBSVM: A library for support vector machines , 2011, TIST.

[40]  Juha Reunanen,et al.  Overfitting in Making Comparisons Between Variable Selection Methods , 2003, J. Mach. Learn. Res..

[41]  David J. Hand,et al.  Multivariate Analysis Of Variance And Repeated Measures , 1987 .

[42]  Gert Rijlaarsdam,et al.  Editorial: Special issue on learning and teaching L2 writing , 2008 .

[43]  Pascal Belin,et al.  Abnormal cortical voice processing in autism , 2004, Nature Neuroscience.

[44]  Nick C Fox,et al.  Automatic classification of MR scans in Alzheimer's disease. , 2008, Brain : a journal of neurology.

[45]  Tom M. Mitchell,et al.  Machine learning classifiers and fMRI: A tutorial overview , 2009, NeuroImage.

[46]  Jennifer H. Pfeifer,et al.  Understanding emotions in others: mirror neuron dysfunction in children with autism spectrum disorders , 2006, Nature Neuroscience.

[47]  M S Buchsbaum,et al.  Anterior cingulate gyrus volume and glucose metabolism in autistic disorder. , 1997, The American journal of psychiatry.

[48]  C. C. Taylor,et al.  Multivariate Analysis of Variance and Repeated Measures: A practical approach for behavioural scientists , 1988 .

[49]  Nathalie Japkowicz,et al.  The class imbalance problem: A systematic study , 2002, Intell. Data Anal..

[50]  C. Frith,et al.  Autism, Asperger syndrome and brain mechanisms for the attribution of mental states to animated shapes. , 2002, Brain : a journal of neurology.

[51]  Shanshan Hong,et al.  Voxel-based morphometry study on brain structure in children with high-functioning autism , 2008, Neuroreport.

[52]  Adam Kowalczyk,et al.  Extreme re-balancing for SVMs: a case study , 2004, SKDD.

[53]  M. Just,et al.  A developmental study of the structural integrity of white matter in autism , 2007, Neuroreport.

[54]  T. Uema,et al.  Abnormal regional cerebral blood flow in childhood autism. , 2000, Brain : a journal of neurology.

[55]  Nitesh V. Chawla,et al.  SMOTE: Synthetic Minority Over-sampling Technique , 2002, J. Artif. Intell. Res..

[56]  H. Akaike A new look at the statistical model identification , 1974 .

[57]  Peg Howland,et al.  Classification using Support Vector Machines with Dimension Reduction , 2004 .

[58]  Jason Weston,et al.  Gene Selection for Cancer Classification using Support Vector Machines , 2002, Machine Learning.

[59]  Nello Cristianini,et al.  Controlling the Sensitivity of Support Vector Machines , 1999 .

[60]  Erin D. Bigler,et al.  Quantitative temporal lobe differences: Autism distinguished from controls using classification and regression tree analysis , 2007, Brain and Development.

[61]  David R. Anderson,et al.  Model selection and multimodel inference : a practical information-theoretic approach , 2003 .

[62]  E. Courchesne,et al.  Reduced size of corpus callosum in autism. , 1995, Archives of neurology.

[63]  Trevor Hastie,et al.  Regularization Paths for Generalized Linear Models via Coordinate Descent. , 2010, Journal of statistical software.

[64]  A. Mayes,et al.  Convergent neuroanatomical and behavioural evidence of an amygdala hypothesis of autism , 2000, Neuroreport.

[65]  Karl J. Friston,et al.  The neuroanatomy of autism: a voxel-based whole brain analysis of structural scans. , 1999, Neuroreport.

[66]  E. Courchesne,et al.  Hypoplasia of cerebellar vermal lobules VI and VII in autism. , 1988, The New England journal of medicine.

[67]  Thomas G. Dietterich Approximate Statistical Tests for Comparing Supervised Classification Learning Algorithms , 1998, Neural Computation.

[68]  Karl J. Friston,et al.  Spatial registration and normalization of images , 1995 .

[69]  Dinggang Shen,et al.  COMPARE: Classification of Morphological Patterns Using Adaptive Regional Elements , 2007, IEEE Transactions on Medical Imaging.

[70]  David J. Hand,et al.  Multivariate Analysis of Variance and Repeated Measures: A practical approach for behavioural scientists , 2011 .