Catalogo Articoli (Spogli Riviste)

OPAC HELP

Titolo:
Penalized discriminant analysis of [O-15]-water PET brain images with prediction error selection of smoothness and regularization hyperparameters
Autore:
Kustra, R; Strother, S;
Indirizzi:
Univ Toronto, Dept Publ Hlth Sci, Toronto, ON, Canada Univ Toronto Toronto ON Canada , Dept Publ Hlth Sci, Toronto, ON, Canada Stanford Univ, Dept Stat, Stanford, CA 94305 USA Stanford Univ Stanford CA USA 94305 iv, Dept Stat, Stanford, CA 94305 USA VA Hosp, PET Imaging, Minneapolis, MN 55417 USA VA Hosp Minneapolis MN USA 55417 , PET Imaging, Minneapolis, MN 55417 USA
Titolo Testata:
IEEE TRANSACTIONS ON MEDICAL IMAGING
fascicolo: 5, volume: 20, anno: 2001,
pagine: 376 - 387
SICI:
0278-0062(200105)20:5<376:PDAO[P>2.0.ZU;2-M
Fonte:
ISI
Lingua:
ENG
Soggetto:
PRINCIPAL-COMPONENT ANALYSIS; POSITRON EMISSION TOMOGRAPHY; SEQUENTIAL FINGER MOVEMENTS; SCALED SUBPROFILE MODEL; CEREBRAL BLOOD-FLOW; FUNCTIONAL CONNECTIVITY; STATISTICAL APPROACH; CROSS-VALIDATION; HIGH-RESOLUTION; LINEAR-MODELS;
Keywords:
bootstrap; discriminant classification; functional neuroimaging; PET; prediction error; regularization;
Tipo documento:
Article
Natura:
Periodico
Settore Disciplinare:
Clinical Medicine
Engineering, Computing & Technology
Citazioni:
62
Recensione:
Indirizzi per estratti:
Indirizzo: Kustra, R Univ Toronto, Dept Publ Hlth Sci, Toronto, ON, Canada Univ Toronto Toronto ON Canada l Hlth Sci, Toronto, ON, Canada
Citazione:
R. Kustra e S. Strother, "Penalized discriminant analysis of [O-15]-water PET brain images with prediction error selection of smoothness and regularization hyperparameters", IEEE MED IM, 20(5), 2001, pp. 376-387

Abstract

We propose a flexible, comprehensive approach for analysis of [O-15]-waterpositron emission tomography (PET) brain images using a penalized version of linear discriminant analysis (PDA), We applied it to scans from 20 subjects (eight scans/subject) performing a finger movement task and analyzed: 1) two classes to obtain a covariance-normalized baseline-activation image, and 2) eight classes for the mean within subject temporal structure which contained baseline-activation and time-dependent changes in a two-dimensional canonical subspace, We imposed spatial smoothness on the resulting image(s) by expanding it in five tenser-product B-spline (TPS) bases of varying smoothness, and further regularized with a ridge-type penalty on the noise covariance matrix. The discrimination approach of PDA provides a probabilistic framework within which prediction error (PE) estimates are derived. We used these to optimize over TPS bases and a ridge hyperparameter (expressed as equivalent degrees of freedom, EDF), We obtained unbiased, low variance PE estimates using modern resampling tools (.632+ Bootstrap and cross validation), and compared PDA of 1) TPS-projected, mean-normalized and unnormalized scans and 2) mean-normalized scans with and without additional presmoothing, By examining the tradeoffs between PE and EDF, as a function of basisselection and image smoothing we demonstrate the utility of PDA, the PE framework, and the relationship between singular value decomposition and smooth TPS bases in the analysis of functional neuroimages.

ASDD Area Sistemi Dipartimentali e Documentali, Università di Bologna, Catalogo delle riviste ed altri periodici
Documento generato il 26/01/20 alle ore 10:21:13