A dynamic graph convolutional neural network framework reveals new insights into connectome dysfunctions in ADHD

Zhao, Kanhao; Duka, Boris; Xie, Hua; Oathes, Desmond J.; Calhoun, Vince D.; Zhang, Yu

doi:10.1016/j.neuroimage.2021.118774

Cited by 83 publications

(68 citation statements)

References 54 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Improved exactness rates accomplished by using an RNN in this experiment demonstrates that these are truly outstanding and precise predictions of progression from MCI to AD. In the future, dynamic graph convolution [47] and multi-view feature learning [48] techniques shall be consider.…”

Section: Discussionmentioning

confidence: 99%

A Long Short-Term Memory Biomarker-Based Prediction Framework for Alzheimer’s Disease

Aqeel

Hassan

Khan

et al. 2022

Sensors

View full text Add to dashboard Cite

The early prediction of Alzheimer’s disease (AD) can be vital for the endurance of patients and establishes as an accommodating and facilitative factor for specialists. The proposed work presents a robotized predictive structure, dependent on machine learning (ML) methods for the forecast of AD. Neuropsychological measures (NM) and magnetic resonance imaging (MRI) biomarkers are deduced and passed on to a recurrent neural network (RNN). In the RNN, we have used long short-term memory (LSTM), and the proposed model will predict the biomarkers (feature vectors) of patients after 6, 12, 21 18, 24, and 36 months. These predicted biomarkers will go through fully connected neural network layers. The NN layers will then predict whether these RNN-predicted biomarkers belong to an AD patient or a patient with a mild cognitive impairment (MCI). The developed methodology has been tried on an openly available informational dataset (ADNI) and accomplished an accuracy of 88.24%, which is superior to the next-best available algorithms.

show abstract

Section: Discussionmentioning

confidence: 99%

A Long Short-Term Memory Biomarker-Based Prediction Framework for Alzheimer’s Disease

Aqeel

Hassan

Khan

et al. 2022

Sensors

View full text Add to dashboard Cite

show abstract

“…We adopted focal loss for the classification: L DGC = −(1 − P r) γ log(P r). As suggested by [23], the focal loss lowers the threshold of accepting a class, inducing a higher recall rate for disorder groups. As KNN is a non-parametric model, trainable parameters come only from node projection, narrowing the parameter searching space and preventing overfitting on the population graph.…”

Section: Population-based Dynamic Graph Classification (Dgc)mentioning

confidence: 92%

“…We collect 596 patients data on AAL90 ROIs to construct individual FC graphs. As implied in [23], incorporating personal characteristic data (PCD) such as age helps to stabilize the metrics. We accordingly keep the three sites KU, KKI, and NYU with no missing values on the 7 PCD features: age, gender, handedness, IQ Measure, Verbal IQ, Performance IQ, and Full4 IQ.…”

Section: Dataset and Experimental Detailsmentioning

confidence: 99%

Contrastive Graph Learning for Population-based fMRI Classification

Wang¹,

Yao²,

Rekik³

2022

Preprint

Self Cite

View full text Add to dashboard Cite

Contrastive self-supervised learning has recently benefited fMRI classification with inductive biases. Its weak label reliance prevents overfitting on small medical datasets and tackles the high intraclass variances. Nonetheless, existing contrastive methods generate resemblant pairs only on pixel-level features of 3D medical images, while the functional connectivity that reveals critical cognitive information is under-explored. Additionally, existing methods predict labels on individual contrastive representation without recognizing neighbouring information in the patient group, whereas interpatient contrast can act as a similarity measure suitable for population-based classification. We hereby proposed contrastive functional connectivity graph learning for population-based fMRI classification. Representations on the functional connectivity graphs are "repelled" for heterogeneous patient pairs meanwhile homogeneous pairs "attract" each other. Then a dynamic population graph that strengthens the connections between similar patients is updated for classification. Experiments on a multi-site dataset ADHD200 validate the superiority of the proposed method on various metrics. We initially visualize the population relationships and exploit potential subtypes.

show abstract

“…In order to solve this problem, graph convolution neural network (GCNN) (Defferrard et al, 2016 ) is introduced to process non-Euclidean data. Zhao et al ( 2022 ) proposed a new dynamic graph convolutional network (dGCN) to learn the potentially important topological information. Song et al ( 2020 ) used dynamic graph convolution network (DGCNN) for the first time in the EEG-based emotion recognition task.…”

Section: Introductionmentioning

confidence: 99%

Linking Multi-Layer Dynamical GCN With Style-Based Recalibration CNN for EEG-Based Emotion Recognition

Bao

Yang

et al. 2022

Front. Neurorobot.

View full text Add to dashboard Cite

Electroencephalography (EEG)-based emotion computing has become one of the research hotspots of human-computer interaction (HCI). However, it is difficult to effectively learn the interactions between brain regions in emotional states by using traditional convolutional neural networks because there is information transmission between neurons, which constitutes the brain network structure. In this paper, we proposed a novel model combining graph convolutional network and convolutional neural network, namely MDGCN-SRCNN, aiming to fully extract features of channel connectivity in different receptive fields and deep layer abstract features to distinguish different emotions. Particularly, we add style-based recalibration module to CNN to extract deep layer features, which can better select features that are highly related to emotion. We conducted two individual experiments on SEED data set and SEED-IV data set, respectively, and the experiments proved the effectiveness of MDGCN-SRCNN model. The recognition accuracy on SEED and SEED-IV is 95.08 and 85.52%, respectively. Our model has better performance than other state-of-art methods. In addition, by visualizing the distribution of different layers features, we prove that the combination of shallow layer and deep layer features can effectively improve the recognition performance. Finally, we verified the important brain regions and the connection relationships between channels for emotion generation by analyzing the connection weights between channels after model learning.

show abstract

A dynamic graph convolutional neural network framework reveals new insights into connectome dysfunctions in ADHD

Cited by 83 publications

References 54 publications

A Long Short-Term Memory Biomarker-Based Prediction Framework for Alzheimer’s Disease

A Long Short-Term Memory Biomarker-Based Prediction Framework for Alzheimer’s Disease

Contrastive Graph Learning for Population-based fMRI Classification

Linking Multi-Layer Dynamical GCN With Style-Based Recalibration CNN for EEG-Based Emotion Recognition

Contact Info

Product

Resources

About