U.S. Department of EnergyOffice of Scientific and Technical Information
Simple Fully Automated Group Classification on Brain fMRI
Abstract
We propose a simple, well grounded classification technique which is suited for group classification on brain fMRI data sets that have high dimensionality, small number of subjects, high noise level, high subject variability, imperfect registration and capture subtle cognitive effects. We propose threshold-split region as a new feature selection method and majority voteas the classification technique. Our method does not require a predefined set of regions of interest. We use average acros ssessions, only one feature perexperimental condition, feature independence assumption, and simple classifiers. The seeming counter-intuitive approach of using a simple design is supported by signal processing and statistical theory. Experimental results in two block design data sets that capture brain function under distinct monetary rewards for cocaine addicted and control subjects, show that our method exhibits increased generalization accuracy compared to commonly used feature selection and classification techniques.
- Authors:
- Publication Date:
- Research Org.:
- Brookhaven National Lab. (BNL), Upton, NY (United States)
- Sponsoring Org.:
- DOE - Office Of Science
- OSTI Identifier:
- 1013544
- Report Number(s):
- BNL-95072-2011-CP
R&D Project: MO-085; KP1602010; TRN: US1102521
- DOE Contract Number:
- DE-AC02-98CH10886
- Resource Type:
- Conference
- Resource Relation:
- Conference: 7th International Sympsoium on Biomedical Imaging: From Nano to Macro (2010 Macro); Rotterdam, Netherlands; 20100414 through 20100417
- Country of Publication:
- United States
- Language:
- English
- Subject:
- 38 RADIATION CHEMISTRY, RADIOCHEMISTRY, AND NUCLEAR CHEMISTRY; ACCURACY; BRAIN; CLASSIFICATION; COCAINE; DESIGN; PROCESSING; pattern classification; fMRI (functional magnetic resonance imaging); Brain imaging; schizophrenia; alzheimer disease; depression; traumatic brain injury
Citation Formats
Honorio, J, Goldstein, R, Honorio, J, Samaras, D, Tomasi, D, and Goldstein, R Z. Simple Fully Automated Group Classification on Brain fMRI. United States: N. p., 2010.
Web.
Honorio, J, Goldstein, R, Honorio, J, Samaras, D, Tomasi, D, & Goldstein, R Z. Simple Fully Automated Group Classification on Brain fMRI. United States.
Honorio, J, Goldstein, R, Honorio, J, Samaras, D, Tomasi, D, and Goldstein, R Z. 2010.
"Simple Fully Automated Group Classification on Brain fMRI". United States. https://www.osti.gov/servlets/purl/1013544.
@article{osti_1013544,
title = {Simple Fully Automated Group Classification on Brain fMRI},
author = {Honorio, J and Goldstein, R and Honorio, J and Samaras, D and Tomasi, D and Goldstein, R Z},
abstractNote = {We propose a simple, well grounded classification technique which is suited for group classification on brain fMRI data sets that have high dimensionality, small number of subjects, high noise level, high subject variability, imperfect registration and capture subtle cognitive effects. We propose threshold-split region as a new feature selection method and majority voteas the classification technique. Our method does not require a predefined set of regions of interest. We use average acros ssessions, only one feature perexperimental condition, feature independence assumption, and simple classifiers. The seeming counter-intuitive approach of using a simple design is supported by signal processing and statistical theory. Experimental results in two block design data sets that capture brain function under distinct monetary rewards for cocaine addicted and control subjects, show that our method exhibits increased generalization accuracy compared to commonly used feature selection and classification techniques.},
doi = {},
url = {https://www.osti.gov/biblio/1013544},
journal = {},
number = ,
volume = ,
place = {United States},
year = {Wed Apr 14 00:00:00 EDT 2010},
month = {Wed Apr 14 00:00:00 EDT 2010}
}