Dimension reduction for individual ICA to decompose FMRI during real-world experiences: Principal component analysis vs. canonical correlation analysis

Research output: Contribution to book/anthology/report/proceedingArticle in proceedingsResearchpeer-review

  • Valeri Tsatsishvili, Department of Mathematical Information Technology, Denmark
  • Fengyu Cong, Department of Mathematical Information Technology
  • ,
  • Tuomas Puoliväli, Department of Mathematical Information Technology, Denmark
  • Vinoo Alluri, Department of Mathematical Information Technology
  • ,
  • Petri Toiviainen, Department of Mathematical Information Technology
  • ,
  • Asoke K. Nandi, Helsingin yliopisto, Denmark
  • Elvira Brattico
  • Tapani Ristaniemi, Department of Mathematical Information Technology

Group independent component analysis (ICA) with special assumptions is often used for analyzing functional magnetic resonance imaging (fMRI) data. Before ICA, dimension reduction is applied to separate signal and noise subspaces. For analyzing noisy fMRI data of individual participants in free-listening to naturalistic and long music, we applied individual ICA and therefore avoided the assumptions of Group ICA. We also compared principal component analysis (PCA) and canonical correlation analysis (CCA) for dimension reduction of such fMRI data. We found interesting brain activity associated with music across majority of participants, and found that PCA and CCA were comparable for dimension reduction.

Original languageEnglish
Title of host publicationESANN 2013 proceedings, 21st European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning
Number of pages6
Publication year2013
Pages137-142
ISBN (print)9782874190810
Publication statusPublished - 2013
Event21st European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, ESANN 2013 - Bruges, Belgium
Duration: 24 Apr 201326 Apr 2013

Conference

Conference21st European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, ESANN 2013
LandBelgium
ByBruges
Periode24/04/201326/04/2013

See relations at Aarhus University Citationformats

ID: 90549268