Jianwen Xie1, Pamela K Douglas2, Ying Nian Wu1, Arthur L Brody2, Ariana E Anderson3. 1. Department of Statistics, University of California, Los Angeles, United States. 2. Department of Psychiatry and Biobehavioral Sciences, University of California, Los Angeles, United States. 3. Department of Statistics, University of California, Los Angeles, United States; Department of Psychiatry and Biobehavioral Sciences, University of California, Los Angeles, United States. Electronic address: arianaanderson@mednet.ucla.edu.
Abstract
BACKGROUND: Brain networks in fMRI are typically identified using spatial independent component analysis (ICA), yet other mathematical constraints provide alternate biologically-plausible frameworks for generating brain networks. Non-negative matrix factorization (NMF) would suppress negative BOLD signal by enforcing positivity. Spatial sparse coding algorithms (L1 Regularized Learning and K-SVD) would impose local specialization and a discouragement of multitasking, where the total observed activity in a single voxel originates from a restricted number of possible brain networks. NEW METHOD: The assumptions of independence, positivity, and sparsity to encode task-related brain networks are compared; the resulting brain networks within scan for different constraints are used as basis functions to encode observed functional activity. These encodings are then decoded using machine learning, by using the time series weights to predict within scan whether a subject is viewing a video, listening to an audio cue, or at rest, in 304 fMRI scans from 51 subjects. RESULTS AND COMPARISON WITH EXISTING METHOD: The sparse coding algorithm of L1 Regularized Learning outperformed 4 variations of ICA (p<0.001) for predicting the task being performed within each scan using artifact-cleaned components. The NMF algorithms, which suppressed negative BOLD signal, had the poorest accuracy compared to the ICA and sparse coding algorithms. Holding constant the effect of the extraction algorithm, encodings using sparser spatial networks (containing more zero-valued voxels) had higher classification accuracy (p<0.001). Lower classification accuracy occurred when the extracted spatial maps contained more CSF regions (p<0.001). CONCLUSION: The success of sparse coding algorithms suggests that algorithms which enforce sparsity, discourage multitasking, and promote local specialization may capture better the underlying source processes than those which allow inexhaustible local processes such as ICA. Negative BOLD signal may capture task-related activations.
BACKGROUND: Brain networks in fMRI are typically identified using spatial independent component analysis (ICA), yet other mathematical constraints provide alternate biologically-plausible frameworks for generating brain networks. Non-negative matrix factorization (NMF) would suppress negative BOLD signal by enforcing positivity. Spatial sparse coding algorithms (L1 Regularized Learning and K-SVD) would impose local specialization and a discouragement of multitasking, where the total observed activity in a single voxel originates from a restricted number of possible brain networks. NEW METHOD: The assumptions of independence, positivity, and sparsity to encode task-related brain networks are compared; the resulting brain networks within scan for different constraints are used as basis functions to encode observed functional activity. These encodings are then decoded using machine learning, by using the time series weights to predict within scan whether a subject is viewing a video, listening to an audio cue, or at rest, in 304 fMRI scans from 51 subjects. RESULTS AND COMPARISON WITH EXISTING METHOD: The sparse coding algorithm of L1 Regularized Learning outperformed 4 variations of ICA (p<0.001) for predicting the task being performed within each scan using artifact-cleaned components. The NMF algorithms, which suppressed negative BOLD signal, had the poorest accuracy compared to the ICA and sparse coding algorithms. Holding constant the effect of the extraction algorithm, encodings using sparser spatial networks (containing more zero-valued voxels) had higher classification accuracy (p<0.001). Lower classification accuracy occurred when the extracted spatial maps contained more CSF regions (p<0.001). CONCLUSION: The success of sparse coding algorithms suggests that algorithms which enforce sparsity, discourage multitasking, and promote local specialization may capture better the underlying source processes than those which allow inexhaustible local processes such as ICA. Negative BOLD signal may capture task-related activations.
Authors: Marta Bianciardi; Masaki Fukunaga; Peter van Gelderen; Jacco A de Zwart; Jeff H Duyn Journal: J Cereb Blood Flow Metab Date: 2010-09-22 Impact factor: 6.200
Authors: Harini Eavani; Roman Filipovych; Christos Davatzikos; Theodore D Satterthwaite; Raquel E Gur; Ruben C Gur Journal: Int Workshop Pattern Recognit Neuroimaging Date: 2012-07-02
Authors: Vince D Calhoun; Vamsi K Potluru; Ronald Phlypo; Rogers F Silva; Barak A Pearlmutter; Arvind Caprihan; Sergey M Plis; Tülay Adalı Journal: PLoS One Date: 2013-08-29 Impact factor: 3.240
Authors: Pamela K Douglas; Edward Lau; Ariana Anderson; Austin Head; Wesley Kerr; Margalit Wollner; Daniel Moyer; Wei Li; Mike Durnhofer; Jennifer Bramen; Mark S Cohen Journal: Front Hum Neurosci Date: 2013-07-31 Impact factor: 3.169