Latent source mining in FMRI data via deep neural network

Heng Huang, Xintao Hu, Junwei Han, Jinglei Lv, Nian Liu, Lei Guo, Tianming Liu

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

22 Scopus citations

Abstract

Independent component analysis (ICA) and its variants have been the dominant methods to the problem of blind source separation (BSS) for functional magnetic resonance imaging (fMRI) data. However, the functional interactions among spatially distributed brain regions and concurrent brain networks deteriorate the basic assumption in ICA-based BSS, that is, the spatial independence of the sources. In this paper, we proposed a novel method for BSS based on recently advanced deep neural network (DNN) algorithm, aiming to detect both internal and functional interaction-induced latent sources simultaneously. We used the motor task fMRI data in the Human Connectome Project (HCP) as a test-bed in the experiments. The results demonstrated the feasibility and effectiveness of the proposed method and its outperformance compared with ICA.

Original languageEnglish
Title of host publication2016 IEEE International Symposium on Biomedical Imaging
Subtitle of host publicationFrom Nano to Macro, ISBI 2016 - Proceedings
PublisherIEEE Computer Society
Pages638-641
Number of pages4
ISBN (Electronic)9781479923502
DOIs
StatePublished - 15 Jun 2016
Event2016 IEEE 13th International Symposium on Biomedical Imaging: From Nano to Macro, ISBI 2016 - Prague, Czech Republic
Duration: 13 Apr 201616 Apr 2016

Publication series

NameProceedings - International Symposium on Biomedical Imaging
Volume2016-June
ISSN (Print)1945-7928
ISSN (Electronic)1945-8452

Conference

Conference2016 IEEE 13th International Symposium on Biomedical Imaging: From Nano to Macro, ISBI 2016
Country/TerritoryCzech Republic
CityPrague
Period13/04/1616/04/16

Keywords

  • blind source separation
  • deep neural network
  • fMRI
  • ICA
  • restricted Boltzmann machine

Fingerprint

Dive into the research topics of 'Latent source mining in FMRI data via deep neural network'. Together they form a unique fingerprint.

Cite this