High‐order resting‐state functional connectivity network for MCI classification

Chen, Xiaobo; Zhang, Han; Gao, Yue; Wee, Chong Yaw; Li, Gang; Shen, Dinggang

doi:10.1002/hbm.23240

Cited by 222 publications

(243 citation statements)

References 68 publications

Supporting

Mentioning

231

Contrasting

Order By: Relevance

“…We compared our method with three state-of-the-art methods: (RAW) where we directly input the raw connectomic brain features, (t-test) where we perform dimensionality reduction using statistical feature selection, and (LLE) where we perform a local linear embedding of the connectomic features to produce a compact and low-dimensional representation of feature vectors. Our method produced the best ASD/NC classification accuracy when using (CC (HON) + CON) features, which significantly outperformed t-test using (CC(HON) + CON) as in [8].…”

Section: Resultsmentioning

confidence: 84%

“…2. Our method produced the best ASD/NC classification accuracy (61.76%) when using (CC(HON) + CON) features, which largely outperformed t-test using (CC(HON) + CON) as in [8].…”

Section: Hj −Hi||2mentioning

confidence: 86%

“…Since both CON and HON feature vectors are high-dimensional, we propose a preliminary dimensionality reduction step through representing each network by a clustering coefficients (CC) feature vector. This will allow us to benchmark our method against the recent connectomic classification framework proposed in [8] where they first concatenated the clustering coefficients of the functional HON (CC(HON)) and CON features (i.e., CC(HON) + CON), then performed t-test for feature selection to train an SVM classifier for Alzheimer's disease diagnosis. We further evaluated all methods using combinations of different feature types: (1) HON, (2) CON, (3) CC(HON), (4) HON + CON, and (5) CC(HON) + CON.…”

Section: Hj −Hi||2mentioning

confidence: 99%

“…The easiest and most commonly employed method for exploring such relationships is simply to concatenate the multiple networks together so that the data from each view is included in the overall set of data for each subject, unaltered [5]. However, recent research on Alzheimer's Disease has found better results when using High Order Networks (HONs) [8]. These are constructed from low-order (e.g.…”

mentioning

confidence: 99%

See 3 more Smart Citations

Autism Spectrum Disorder Diagnosis Using Sparse Graph Embedding of Morphological Brain Networks

Morris

Rekik

2017

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder involving a complex cognitive impairment that can be difficult to diagnose early enough. Much work has therefore been done investigating the use of machine-learning techniques on functional and structural connectivity networks for ASD diagnosis. However, networks based on the morphology of the brain have yet to be similarly investigated, despite research findings that morphological features, such as cortical thickness, are affected by ASD. In this paper, we first propose modelling morphological brain connectivity (or graph) using a set of cortical attributes, each encoding a unique aspect of cortical morphology. However, it can be difficult to capture for each subject the complex pattern of relationships between morphological brain graphs, where each may be affected simultaneously or independently by ASD. In order to solve this problem, we therefore also propose the use of high-order networks which can better capture these relationships. Further, since ASD and normal control (NC) high-dimensional connectomic data might lie in different manifolds, we aim to find a low-dimensional representation of the data which captures the intrinsic dimensions of the underlying connectomic manifolds, thereby allowing better learning by linear classifiers. Hence, we propose the use of sparse graph embedding (SGE) method, which allows us to distinguish between data points drawn from different manifolds, even when they are too close to one another. SGE learns a similarity matrix of the connectomic data graph, which then is used to embed the high-dimensional connectomic features into a low-dimensional space that preserves the locality of the original data. Our ASD/NC classification results outperformed several state-of-the-art methods including statistical feature selection, and local linear embedding methods.

show abstract

Section: Resultsmentioning

confidence: 84%

“…2. Our method produced the best ASD/NC classification accuracy (61.76%) when using (CC(HON) + CON) features, which largely outperformed t-test using (CC(HON) + CON) as in [8].…”

Section: Hj −Hi||2mentioning

confidence: 86%

Section: Hj −Hi||2mentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

Autism Spectrum Disorder Diagnosis Using Sparse Graph Embedding of Morphological Brain Networks

Morris

Rekik

2017

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…On the other hand, [4] computed sparse temporal networks using sliding-window approach over a time series of restingstate functional MRI. [5] extended this work by additionally considering the high-order correlation between different pairs of brain regions. By combining low-order with high-order correlations, they further improved the classification accuracy of eMCI/NC.…”

Section: Introductionmentioning

confidence: 99%

Pairing-based Ensemble Classifier Learning using Convolutional Brain Multiplexes and Multi-view Brain Networks for Early Dementia Diagnosis

Lisowska

Rekik

2017

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. The majority of works using brain connectomics for dementia diagnosis heavily relied on using structural (diffusion MRI) and functional brain connectivity (functional MRI). However, how early dementia affects the morphology of the cortical surface remains poorly understood. In this paper, we first introduce multi-view morphological brain network architecture which stacks multiple networks, each quantifying a cortical attribute (e.g., thickness). Second, to model the relationship between brain views, we propose a subject-specific convolutional brain multiplex composed of intra-layers (brain views) and inter-layers between them by convolving two consecutive views. By reordering the intra-layers, we generate different multiplexes for each subject. Third, to distinguish demented brains from healthy ones, we propose a pairing-based ensemble classifier learning strategy, which projects each pair of brain multiplex sets onto a low-dimensional space where they are fused, then classified. Our framework achieved the best classification results for the right hemisphere 90.8% and the left hemisphere 89.5%.

show abstract

Functional connectivity between white matter and gray matter based on fMRI for Alzheimer's disease classification

Zhao

Ding

et al. 2019

Brain and Behavior

View full text Add to dashboard Cite

Introduction Alzheimer's disease (AD) is a chronic neurodegenerative disease that generally starts slowly and leads to deterioration over time. Finding biomarkers more effective to predict AD transition is important for clinical medicine. And current research indicated that the lesion regions occur in both gray matter (GM) and white matter (WM). Methods This paper extracted BOLD time series from WM and GM, combined WM and GM together for analysis, constructed functional connectivity (FC) of static (sWGFC) and dynamic (dWGFC) between WM and GM, as well as static (sGFC) and dynamic (dGFC) FC within GM in order to evaluate the methods and areas most useful as feature sets for distinguishing NC from AD. These features will be evaluated using support vector machine (SVM) classifiers. Results The FC constructed by WM BOLD time series based on fMRI showed widely differences between the AD group and NC group. In terms of the results of the classification, the performance of feature subsets selected from sWGFC was better than sGFC, and the performance of feature subsets selected from dWGFC was better than dGFC. Overall, the feature subsets selected from dWGFC was the best. Conclusion These results indicated that there is a wide range of disconnection between WM and GM in AD, and association between WM and GM based on fMRI only is an effective strategy, and the FC between WM and GM could be a potential biomarker in the process of cognitive impairment and AD.

show abstract

High‐order resting‐state functional connectivity network for MCI classification

Cited by 222 publications

References 68 publications

Autism Spectrum Disorder Diagnosis Using Sparse Graph Embedding of Morphological Brain Networks

Autism Spectrum Disorder Diagnosis Using Sparse Graph Embedding of Morphological Brain Networks

Pairing-based Ensemble Classifier Learning using Convolutional Brain Multiplexes and Multi-view Brain Networks for Early Dementia Diagnosis

Functional connectivity between white matter and gray matter based on fMRI for Alzheimer's disease classification

Contact Info

Product

Resources

About