Robust brain parcellation using sparse representation on resting-state fMRI

Caspers, Svenja; Fan, Lingzhong; Fan, Yong; Song, Ming; Liu, Cirong; Mo, Yin; Roski, Christian; Eickhoff, Simon B.; Amunts, Katrin; Jiang, Tianzi

doi:10.1007/s00429-014-0874-x

Cited by 27 publications

(27 citation statements)

References 69 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For GMVCorr (Figure c), we calculated the Pearson correlation between the GMV of each hippocampal ROI voxel (from the 1.5 × 1.5 × 1.5 mm 3 set) and those from each nonhippocampal voxel in the rest of the brain (from the 3 × 3 × 3 mm 3 set; downsampling was used to obtain sufficient spatial resolution for each ROI region and to meet reasonable computational and memory requirements) across all 198 subjects. The coefficient matrix C ∈ R v × w of all ROI voxels was calculated ( v and w were the number of voxels over the ROI region and nonhippocampal regions), and a symmetric nonnegative similarity matrix W (Elhamifar & Vidal, ; Ge et al, ) was constructed from the coefficient matrix: W = | C · E − 1 · C T |∈ R v × v (Zhang et al, ), where E was the diagonal matrix saving the column summations of the coefficient matrix C and T denoted transposition. The similarity matrix W was then fed into a spectral clustering algorithm (Ge et al, ; Ng, Jordan, & Weiss, ) to generate the final parcellations.…”

Section: Methodsmentioning

confidence: 99%

Parcellation of the human hippocampus based on gray matter volume covariance: Replicable results on healthy young adults

Kot

Liu

et al. 2019

Human Brain Mapping

View full text Add to dashboard Cite

The hippocampus is a key brain region that participates in a range of cognitive and affective functions, and is involved in the etiopathogenesis of numerous neuropsychiatric disorders. The structural complexity and functional diversity of the hippocampus suggest the existence of structural and functional subdivisions within this structure. For the first time, we parcellated the human hippocampus with two independent data sets, each of which consisted of 198 T1‐weighted structural magnetic resonance imaging (sMRI) images of healthy young subjects. The method was based on gray matter volume (GMV) covariance, which was quantified by a bivariate voxel‐to‐voxel linear correlation approach, as well as a multivariate masked independent component analysis approach. We subsequently interrogated the relationship between the GMV covariance patterns and the functional connectivity patterns of the hippocampal subregions using sMRI and resting‐state functional MRI (fMRI) data from the same participants. Seven distinct GMV covariance‐based subregions were identified for bilateral hippocampi, with robust reproducibility across the two data sets. We further demonstrated that the structural covariance patterns of the hippocampal subregions had a correspondence with the intrinsic functional connectivity patterns of these subregions. Together, our results provide a topographical configuration of the hippocampus with converging structural and functional support. The resulting subregions may improve our understanding of the hippocampal connectivity and functions at a subregional level, which provides useful parcellations and masks for future neuroscience and clinical research on the structural and/or functional connectivity of the hippocampus.

show abstract

Section: Methodsmentioning

confidence: 99%

Parcellation of the human hippocampus based on gray matter volume covariance: Replicable results on healthy young adults

Kot

Liu

et al. 2019

Human Brain Mapping

View full text Add to dashboard Cite

show abstract

“…This defect restricts the use of this technique. Both a whole brain connectivity strategy and the local covariance have been used in resting state fMRI-based parcellations (Yeo et al, 2011; Zhang et al, 2014). Since our goal was to obtain brain function-structure mapping, we were obliged to dig deeply into the task-fMRI data.…”

Section: Discussionmentioning

confidence: 99%

“…Although the covariance of resting state signal fluctuations is conceptually different from these methods, some researchers have directly used covariance to parcellate brain areas (Zhang et al, 2014). Similarly, another study used the Brainmap database to investigate the covariance within the activation pattern rather than focusing on co-activations (Smith et al, 2009).…”

Section: Introductionmentioning

confidence: 99%

Identifying functional subdivisions in the human brain using meta-analytic activation modeling-based parcellation

et al. 2016

Self Cite

View full text Add to dashboard Cite

Parcellation of the human brain into fine-grained units by grouping voxels into distinct clusters has been an effective approach for delineating specific brain regions and their subregions. Published neuroimaging studies employing coordinate-based meta-analyses have shown that the activation foci and their corresponding behavioral categories may contain useful information about the anatomical–functional organization of brain regions. Inspired by these developments, we proposed a new parcellation scheme called meta-analytic activation modeling-based parcellation (MAMP) that uses meta-analytically obtained information. The raw meta data, including the experiments and the reported activation coordinates related to a brain region of interest, were acquired from the Brainmap database. Using this data, we first obtained the “modeled activation” pattern by modeling the voxel-wise activation probability given spatial uncertainty for each experiment that featured at least one focus within the region of interest. Then, we processed these “modeled activation” patterns across the experiments with a K-means clustering algorithm to group the voxels into different subregions. In order to verify the reliability of the method, we employed our method to parcellate the amygdala and the left Brodmann area 44 (BA44). The parcellation results were quite consistent with previous cytoarchitectonic and in vivo neuroimaging findings. Therefore, the MAMP proposed in the current study could be a useful complement to other methods for uncovering the functional organization of the human brain.

show abstract

“…Secondly, both noise and redundancy of the neuroimaging and genetic data could reduce the rank of the data, which may make it difficult to obtain robust machine learning models from such data (Vounou et al 2012; Greenlaw et al 2017; Lu et al 2017). To alleviate the “curse of dimensionality” problem, data reduction techniques have been widely adopted in imaging-genetics studies, including linear subspace analysis methods and supervoxel methods (Fan et al 2005, 2007; Zhang et al 2015). In particular, linear subspace learning methods, such as principle component analysis (PCA), have been used to project voxelwise measures into a small number of components, and the supervoxel methods parcellate the brain into regions of interest (ROIs) by adopting anatomical atlases or clustering image voxels.…”

Section: Introductionmentioning

confidence: 99%

A Robust Reduced Rank Graph Regression Method for Neuroimaging Genetic Analysis

2018

Self Cite

View full text Add to dashboard Cite

To characterize associations between genetic and neuroimaging data, a variety of analytic methods have been proposed in neuroimaging genetic studies. These methods have achieved promising performance by taking into account inherent correlation in either the neuroimaging data or the genetic data alone. In this study, we propose a novel robust reduced rank graph regression based method in a linear regression framework by considering correlations inherent in neuroimaging data and genetic data jointly. Particularly, we model the association analysis problem in a reduced rank regression framework with the genetic data as a feature matrix and the neuroimaging data as a response matrix by jointly considering correlations among the neuroimaging data as well as correlations between the genetic data and the neuroimaging data. A new graph representation of genetic data is adopted to exploit their inherent correlations, in addition to robust loss functions for both the regression and the data representation tasks, and a square-root-operator applied to the robust loss functions for achieving adaptive sample weighting. The resulting optimization problem is solved using an iterative optimization method whose convergence has been theoretically proved. Experimental results on the Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset have demonstrated that our method could achieve competitive performance in terms of regression performance between brain structural measures and the Single Nucleotide Polymorphisms (SNPs), compared with state-of-the-art alternative methods.

show abstract

Robust brain parcellation using sparse representation on resting-state fMRI

Cited by 27 publications

References 69 publications

Parcellation of the human hippocampus based on gray matter volume covariance: Replicable results on healthy young adults

Parcellation of the human hippocampus based on gray matter volume covariance: Replicable results on healthy young adults

Identifying functional subdivisions in the human brain using meta-analytic activation modeling-based parcellation

A Robust Reduced Rank Graph Regression Method for Neuroimaging Genetic Analysis

Contact Info

Product

Resources

About