Premium
Unmixing functional magnetic resonance imaging data using matrix factorization
Author(s) -
Khaliq Amir A.,
Qureshi Ijaz M.,
Shah Jawad A.
Publication year - 2012
Publication title -
international journal of imaging systems and technology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.359
H-Index - 47
eISSN - 1098-1098
pISSN - 0899-9457
DOI - 10.1002/ima.22022
Subject(s) - non negative matrix factorization , principal component analysis , matrix decomposition , functional magnetic resonance imaging , computer science , pattern recognition (psychology) , independent component analysis , voxel , artificial intelligence , physics , eigenvalues and eigenvectors , quantum mechanics , neuroscience , biology
Abstract Functional magnetic resonance imaging (fMRI) data is processed by different techniques for detection of activated voxels including principal component analysis (PCA), independent component analysis (ICA), non‐negative matrix factorization (NMF), and so on. In this work, a modified version of NMF method is proposed in which data is not supposed to be non‐negative. The proposed scheme is applied to synthetic fMRI data along with NMF conventional method. The results of the proposed scheme show that it is not only computationally efficient but also has good quality results as compared to that of NMF in terms of average correlation. Finally, proposed method is applied to monkey's fMRI data, and the results are compared with that of NMF and ICA. © 2012 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 22, 195–199, 2012