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

Unsupervised analysis of fMRI data using kernel canonical correlation.

David R Hardoon1, Janaina Mourão-Miranda, Michael Brammer

  • 1The Centre for Computational Statistics and Machine Learning, Department of Computer Science, University College London, UK. D.Hardoon@cs.ucl.ac.uk

Neuroimage
|August 10, 2007
PubMed
Summary
This summary is machine-generated.

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

Cardiometabolic health and physical robustness map onto distinct patterns of brain structure and neurotransmitter systems.

PLoS biology·2025
Same author

Evaluating analytic strategies to obtain high-resolution, vertex-level measures of cortical neuroanatomy in children in low- and middle-income countries.

Communications biology·2025
Same author

Contrastive learning of T cell receptor representations.

Cell systems·2025
Same author

Effects of cannabidiol on symptoms in people at clinical high risk for psychosis.

World psychiatry : official journal of the World Psychiatric Association (WPA)·2024
Same author

Hippocampal Glutamate, Resting Perfusion and the Effects of Cannabidiol in Psychosis Risk.

Schizophrenia bulletin open·2024
Same author

Voxelwise Multivariate Analysis of Brain-Psychosocial Associations in Adolescents Reveals 6 Latent Dimensions of Cognition and Psychopathology.

Biological psychiatry. Cognitive neuroscience and neuroimaging·2024
Same journal

Segmentation of the parasagittal dura mater on multi-center 3D-FLAIR MRI.

NeuroImage·2026
Same journal

Spatial frequency channels implement a mental ruler in spatial vision.

NeuroImage·2026
Same journal

Exploring the Link Between Intravoxel Incoherent Motion Measured Brain Diffusivity During Wakefulness and Sleep Macrostructure in the Elderly.

NeuroImage·2026
Same journal

Closed-loop adaptation of transcranial magnetic stimulation intensity with electroencephalography feedback.

NeuroImage·2026
Same journal

Volumetric postmortem MRI of the medial temporal lobe in Alzheimer's disease and related disorders: methodological advances and implications for in vivo biomarker development.

NeuroImage·2026
Same journal

Neural responses to equity and inequity when receiving vicarious rewards for self and charity during adolescence.

NeuroImage·2026
See all related articles

We developed a new unsupervised fMRI analysis method, Kernel Canonical Correlation Analysis (KCCA), which uses detailed stimulus features instead of simple labels. KCCA shows comparable accuracy to Support Vector Machines (SVM) while identifying key brain regions, particularly in the visual cortex.

Area of Science:

  • Neuroimaging
  • Machine Learning
  • Data Analysis

Background:

  • Supervised learning methods, like Support Vector Machines (SVM), are increasingly used for functional Magnetic Resonance Imaging (fMRI) analysis.
  • These methods typically associate imaging data with simple categorical labels representing experimental conditions.

Purpose of the Study:

  • Introduce a novel unsupervised fMRI analysis method, Kernel Canonical Correlation Analysis (KCCA).
  • Compare KCCA with SVM using fMRI data from emotionally salient stimuli.
  • Evaluate KCCA's ability to identify task-discriminative brain regions without prior categorical labels.

Main Methods:

  • Kernel Canonical Correlation Analysis (KCCA) was employed as an unsupervised learning method.
  • KCCA utilizes detailed feature vectors for each stimulus, differing from SVM's categorical labels.

Related Experiment Videos

  • Both KCCA and SVM were trained and tested on fMRI data of responses to pleasant and unpleasant stimuli.

    • Main Results:

    • KCCA and SVM achieved similar classification accuracies in discriminating between stimuli.
    • KCCA successfully identified important brain regions for task discrimination, primarily in the visual cortex.
    • KCCA identified these regions independently of categorical task labels, deriving stimulus category from image features.

    Conclusions:

    • KCCA offers a viable unsupervised alternative for fMRI analysis.
    • KCCA can effectively identify task-relevant brain regions, complementing supervised methods.
    • The method's ability to use detailed stimulus features enhances its potential in neuroimaging research.