Detecting outliers in high-dimensional neuroimaging datasets with robust covariance estimators

• Published in: Medical image analysis Vol. 16; no. 7; pp. 1359 - 1370
• Main Authors: Fritsch, Virgile, Varoquaux, Gaël, Thyreau, Benjamin, Poline, Jean-Baptiste, Thirion, Bertrand
• Format: Journal Article
• Language: English
• Published: Netherlands Elsevier B.V 01.10.2012; Elsevier
• Subjects: Algorithms; Applications; Artificial Intelligence; Bioengineering; Computer Science; High-dimension; Humans; Image Enhancement - methods; Image Interpretation, Computer-Assisted - methods; Imaging; Information Storage and Retrieval - methods; Life Sciences; Medical Imaging; Minimum covariance determinant; Neuroimaging; Neuroimaging - methods; Outlier detection; Pattern Recognition, Automated - methods; Reproducibility of Results; Robust estimation; Sensitivity and Specificity; Statistics; Subtraction Technique; Minimum covariance determinant; Neuroimaging; Outlier detection; Robust estimation; High-dimension; fMRI; high-dimension; robust estimation; regularization; outlier detection; minimum covariance determinant; neuroimaging
• ISSN: 1361-8415; 1361-8423; 1361-8423
• DOI: 10.1016/j.media.2012.05.002

[Display omitted] ► Multivariate outlier detection can screen deviant acquisitions in medical imaging. ► For high-dimensional data state-of-the-art robust estimator must be regularized. ► We study different approaches to regularization. ► MCD with random projections gives good control on outliers ev...