Statistical inference and classification for neuroimaging
Work at Parietal on statistical analysis for multi subject analysis of neuroimaging data, and the comparison of neuroimaging data with genetic or behavioural information.
Discriminating Between Populations of Subjects Based on FMRI Data Using Sparse Features Selection and SRDA Classifier
Work of Damon et al, functional workshop MICCAI 2008
Outlier detection on an fMRI contrast
R-MCD-based Mahalanobis distances of a small sample. The higher the Mahalanobis distance, the higher the probability for an observation to be tagged as outlying. Points in red are outliers subjects according to the whole population.
HiDiNim: High-dimensional Neuroimaging -- Statistical Models of Brain Variability observed in Neuroimaging
In this work, we propose to investigate the statistical structure of large populations observed in neuroimaging. In particular, we will investigate the use of region-level averages of brain activity, that we plan to co-analyse with genetic and behavioral information, in order to understand the sources of the observed variability.
Outlier detection accuracy of RMCD
Outlier detection accuracy for MCD- and R-MCD-based outliers detection methods, measured by Area Under Curve values. 40% multivariate outliers are generated (α = 5, κ(Σ) = 1000). The R-MCD-based method keeps an AUC of 0.70 up to n = 3.5 (not shown) while the MCD-based method breaks down.
Home
