Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Strategies for reducing large fMRI data sets for independent component analysis.

Ze Wang1, Jiongjiong Wang, Vince Calhoun

  • 1Center for Functional Neuroimaging, University of Pennsylvania, Philadelphia, 19104, USA. zewang@mail.med.upenn.edu

Magnetic Resonance Imaging
|June 1, 2006
PubMed
Summary

Principal component analysis (PCA) struggles with large fMRI data for temporal ICA (tICA). A new Cascade Recursive Least Squared (CRLS) network method offers efficient data reduction for both spatial and temporal ICA, overcoming computational and memory limitations.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Diurnal Variations and Test-Retest Reliability of Resting-State Functional MRI Metrics.

Human brain mapping·2026
Same author

Applications of humanized mice in <i>Mycobacterium tuberculosis</i> infections (Review).

Molecular medicine reports·2026
Same author

Decoding the adenosinergic paradox of chronic sleep restriction.

Sleep·2026
Same author

Retraction notice to "Test-retest reliability of coupling between cerebrospinal fluid flow and global brain activity after normal sleep and sleep deprivation" [NeuroImage 309 (2025) 121097].

NeuroImage·2026
Same author

Network analysis of childhood trauma and meaning in life in adolescents with and without depression.

BMC psychology·2026
Same author

T-cell epitope supplementation enhances cross-protection against H7N9 AIV in chickens via upregulated HA epitope-specific T cell and CD4 T cell-mediated help.

Journal of advanced research·2026

Area of Science:

  • Neuroimaging
  • Computational Neuroscience
  • Signal Processing

Background:

  • Principal Component Analysis (PCA) is crucial for data reduction in Independent Component Analysis (ICA).
  • Traditional PCA using eigenvector decomposition on the covariance matrix is computationally intractable for high-dimensional fMRI data, especially in temporal ICA (tICA).
  • Existing methods face challenges with large datasets, high spatial dimensions, and computational intensity.

Purpose of the Study:

  • To address the limitations of conventional PCA in fMRI data analysis for ICA.
  • To introduce practical data reduction methods for both spatial ICA (sICA) and temporal ICA (tICA).
  • To present a novel Cascade Recursive Least Squared (CRLS) network for efficient PCA decomposition.

Main Methods:

Related Experiment Videos

  • Proposed two data reduction methods: calculating tICA PCs from sICA PCs, and PCA decomposition via a CRLS network.
  • CRLS-PCA extracts Principal Components (PCs) directly from raw data, bypassing the need for covariance matrix calculation.
  • CRLS-PCA allows for arbitrary termination of PC extraction and is optimized for large datasets exceeding machine memory.

    • Main Results:

    • The CRLS-PCA method provides a uniform data reduction solution for both sICA and tICA.
    • CRLS-PCA demonstrates improved efficiency in computational expense and memory usage compared to conventional PCA.
    • Evaluation on real fMRI data confirmed the precision, computational efficiency, and memory savings of the CRLS-PCA method.

    Conclusions:

    • The CRLS-PCA network offers a scalable and efficient solution for PCA-based data reduction in fMRI analysis.
    • This method overcomes the computational and memory bottlenecks associated with traditional PCA for large-scale neuroimaging data.
    • CRLS-PCA facilitates more tractable analysis of both spatial and temporal components in fMRI data.