DocumentCode
393920
Title
Characterization of the correlation, SPM, ICA and specified-resolution wavelet methods in analysis of fMRI study with visual task
Author
Chen, C.L. ; Wu, T.H. ; Wu, Y.T. ; Hsieh, J.C. ; Huang, Y.H. ; Lee, J.S.
Author_Institution
Inst. of Radiol. Sci., Nat. Yang-Ming Univ., Taipei, Taiwan
fYear
2003
fDate
20-22 March 2003
Firstpage
557
Lastpage
560
Abstract
An inter-comparison study among the correlation, the SPM, the ICA and the specified-resolution wavelet analytical methods on the performance of finding fMRI task-related brain activations in a visual task is presented. Each of the methods has its own unique pre-assumptions and contributions, and potential variations or uncertainties underlying these methods are very different from one another. One should comprehend the assumptions and theories of each method in depth, and optimize the appropriate techniques for functional studies. The advantages and disadvantages of the analytical methods were systemically characterized and compared, and from our study results, the ICA method seemed to be a more suitable choice for a visual task in fMRI; the shapes of its task-related time courses in the common region shared by all techniques were much more coherent with a predefined reference function.
Keywords
biomedical MRI; data analysis; statistical analysis; vision; wavelet transforms; ICA; SPM; correlation method; fMRI study; fMRI task-related brain activations; independent component analysis; inter-comparison study; predefined reference function; specified-resolution wavelet methods; statistical parametric mapping; visual task; Brain; Correlation; Data analysis; Image analysis; Independent component analysis; Magnetic resonance imaging; Scanning probe microscopy; Signal design; Uncertainty; Wavelet analysis;
fLanguage
English
Publisher
ieee
Conference_Titel
Neural Engineering, 2003. Conference Proceedings. First International IEEE EMBS Conference on
Print_ISBN
0-7803-7579-3
Type
conf
DOI
10.1109/CNE.2003.1196887
Filename
1196887
Link To Document