Sparse SPM: Group Sparse-dictionary learning in SPM framework for resting-state functional connectivity MRI analysis

Lee, Young-Beom; Lee, Jeonghyeon; Tak, Sungho; Lee, K.; Na, Duk L.; Seo, Sang Won; Jeong, Yong Hoon; Ye, Jong Chul

doi:10.1016/j.neuroimage.2015.10.081

Cited by 43 publications

(42 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To address the limitations of conventional approaches, dictionary learning‐based methods have been widely used for analyzing functional networks [Lee et al, ; Lv et al, 2015a, 2015b]. Unlike conventional approaches such as PCA and ICA, dictionary learning does not impose that source signals are orthogonal or independent, allowing more flexibility in adapting the representation to the data [Mairal et al, ].…”

Section: Introductionmentioning

confidence: 99%

Multiple functional networks modeling for autism spectrum disorder diagnosis

Kam

Suk

Lee

2017

Human Brain Mapping

View full text Add to dashboard Cite

Despite countless studies on autism spectrum disorder (ASD), diagnosis relies on specific behavioral criteria and neuroimaging biomarkers for the disorder are still relatively scarce and irrelevant for diagnostic workup. Many researchers have focused on functional networks of brain activities using resting-state functional magnetic resonance imaging (rsfMRI) to diagnose brain diseases, including ASD. Although some existing methods are able to reveal the abnormalities in functional networks, they are either highly dependent on prior assumptions for modeling these networks or do not focus on latent functional connectivities (FCs) by considering discriminative relations among FCs in a nonlinear way. In this article, we propose a novel framework to model multiple networks of rsfMRI with data-driven approaches. Specifically, we construct large-scale functional networks with hierarchical clustering and find discriminative connectivity patterns between ASD and normal controls (NC). We then learn features and classifiers for each cluster through discriminative restricted Boltzmann machines (DRBMs). In the testing phase, each DRBM determines whether a test sample is ASD or NC, based on which we make a final decision with a majority voting strategy. We assess the diagnostic performance of the proposed method using public datasets and describe the effectiveness of our method by comparing it to competing methods. We also rigorously analyze FCs learned by DRBMs on each cluster and discover dominant FCs that play a major role in discriminating between ASD and NC. Hum Brain Mapp 38:5804-5821, 2017. © 2017 Wiley Periodicals, Inc.

show abstract

Section: Introductionmentioning

confidence: 99%

Multiple functional networks modeling for autism spectrum disorder diagnosis

Kam

Suk

Lee

2017

Human Brain Mapping

View full text Add to dashboard Cite

show abstract

“…This technology has been widely used to identify biomarkers of AD based on brain network alterations (Wang et al, 2007; Agosta et al, 2012; Sui et al, 2015). Seed-based approaches (Fox et al, 2009), independent components analysis (ICA) based approaches (Lee et al, 2015) and graph theory (Zhang et al, 2011) have been the three primary methods used in the study of resting-state functional connectivity (FC) in the brain. The seed-based approach involves predefining a region of interest (ROI) and extracting the BOLD signal from it; then a map of FC is obtained by calculating the cross-correlation between the time series extracted from the seed ROI and all other voxels in the brain.…”

Section: Introductionmentioning

confidence: 99%

Investigating Focal Connectivity Deficits in Alzheimer's Disease Using Directional Brain Networks Derived from Resting-State fMRI

Zhao

Rangaprakash

Venkataraman

et al. 2017

Front. Aging Neurosci.

View full text Add to dashboard Cite

Connectivity analysis of resting-state fMRI has been widely used to identify biomarkers of Alzheimer's disease (AD) based on brain network aberrations. However, it is not straightforward to interpret such connectivity results since our understanding of brain functioning relies on regional properties (activations and morphometric changes) more than connections. Further, from an interventional standpoint, it is easier to modulate the activity of regions (using brain stimulation, neurofeedback, etc.) rather than connections. Therefore, we employed a novel approach for identifying focal directed connectivity deficits in AD compared to healthy controls. In brief, we present a model of directed connectivity (using Granger causality) that characterizes the coupling among different regions in healthy controls and Alzheimer's disease. We then characterized group differences using a (between-subject) generative model of pathology, which generates latent connectivity variables that best explain the (within-subject) directed connectivity. Crucially, our generative model at the second (between-subject) level explains connectivity in terms of local or regionally specific abnormalities. This allows one to explain disconnections among multiple regions in terms of regionally specific pathology; thereby offering a target for therapeutic intervention. Two foci were identified, locus coeruleus in the brain stem and right orbitofrontal cortex. Corresponding disrupted connectivity network associated with the foci showed that the brainstem is the critical focus of disruption in AD. We further partitioned the aberrant connectomic network into four unique sub-networks, which likely leads to symptoms commonly observed in AD. Our findings suggest that fMRI studies of AD, which have been largely cortico-centric, could in future investigate the role of brain stem in AD.

show abstract

“…In the current field of functional neuroimaging research, one of the most effective approaches for fMRI data analysis is the functional network decomposition based on Dictionary Learning methods [1,2]. Dictionary learning derives a set of vectors that sparsely code the input fMRI data.…”

Section: Introductionmentioning

confidence: 99%

“…The resulting dictionaries and sparsenessconstraint loading coefficients respectively characterize the underlying temporal and spatial distribution patterns of the atomic functional networks over the whole brain. Both individual [3] and group-wise dictionary learning studies [1] on fMRI data have been conducted on task and restingstate fMRI data. It has been shown in various studies that specific functional network alterations could help identify brain disorders (e.g.…”

Section: Introductionmentioning

confidence: 99%

“…Following the previous success in using dictionary learning for functional network decomposition [1,2], in this work we developed a novel distributed rank-1 dictionary learning (D-r1DL) framework [19], using the Apache Spark platform. All code presented in this work can be found in the Github repository https://github.com/quinngroup/dr1dl-pyspark Dictionary learning has been in the center of attention of researchers in variety of disciplines, but less effort has been made to create new algorithms using it.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Distributed rank-1 dictionary learning: Towards fast and scalable solutions for fMRI big data analytics

Makkie

Liu

et al. 2016

2016 IEEE International Conference on Big Data (Big Data)

View full text Add to dashboard Cite

Abstract_The use of functional brain imaging for research and diagnosis has benefitted greatly from the recent advancements in neuroimaging technologies, as well as the explosive growth in size and availability of fMRI data. While it has been shown in literature that using multiple and large scale fMRI datasets can improve reproducibility and lead to new discoveries, the computational and informatics systems supporting the analysis and visualization of such fMRI big data are extremely limited and largely under-discussed. We propose to address these shortcomings in this work, based on previous success in using dictionary learning method for functional network decomposition studies on fMRI data. We presented a distributed dictionary learning framework based on rank-1 matrix decomposition with sparseness constraint (D-r1DL framework). The framework was implemented using the Spark distributed computing engine and deployed on three different processing units: an in-house server, inhouse high performance clusters, and the Amazon Elastic Compute Cloud (EC2) service. The whole analysis pipeline was integrated with our neuroinformatics system for data management, user input/output, and real-time visualization. Performance and accuracy of D-r1DL on both individual and group-wise fMRI Human Connectome Project (HCP) dataset shows that the proposed framework is highly scalable. The resulting group-wise functional network decompositions are highly accurate, and the fast processing time confirm this claim. In addition, D-r1DL can provide real-time user feedback and results visualization which are vital for largescale data analysis.

show abstract

Sparse SPM: Group Sparse-dictionary learning in SPM framework for resting-state functional connectivity MRI analysis

Cited by 43 publications

References 32 publications

Multiple functional networks modeling for autism spectrum disorder diagnosis

Multiple functional networks modeling for autism spectrum disorder diagnosis

Investigating Focal Connectivity Deficits in Alzheimer's Disease Using Directional Brain Networks Derived from Resting-State fMRI

Distributed rank-1 dictionary learning: Towards fast and scalable solutions for fMRI big data analytics

Contact Info

Product

Resources

About