Classify EEG and Reveal Latent Graph Structure with Spatio-Temporal Graph Convolutional Neural Network

Li, Xiaoyü; Qian, Buyue; Wei, Jishang; An, Lihui; Liu, Xuan; Zheng, Qinghua

doi:10.1109/icdm.2019.00049

Cited by 18 publications

(20 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…During the training process, the weight matrices learn the dynamic latent graph structure. Such approaches have been proposed for MDD [ 60 ], ASD [ 36 ] and emotion recognition [ 6 , 185 ]. Furthermore, the model should learn both local and global spatial information; however,, in the majority of surveyed works, each node is traditionally connected only to its spatially closest neighbours, which leads to a very limited information exchange between distant nodes.…”

Section: Research Challenges and Future Directionsmentioning

confidence: 99%

Graph-Based Deep Learning for Medical Diagnosis and Analysis: Past, Present and Future

Ahmedt-Aristizabal

Armin

Denman

et al. 2021

Sensors

136

View full text Add to dashboard Cite

With the advances of data-driven machine learning research, a wide variety of prediction problems have been tackled. It has become critical to explore how machine learning and specifically deep learning methods can be exploited to analyse healthcare data. A major limitation of existing methods has been the focus on grid-like data; however, the structure of physiological recordings are often irregular and unordered, which makes it difficult to conceptualise them as a matrix. As such, graph neural networks have attracted significant attention by exploiting implicit information that resides in a biological system, with interacting nodes connected by edges whose weights can be determined by either temporal associations or anatomical junctions. In this survey, we thoroughly review the different types of graph architectures and their applications in healthcare. We provide an overview of these methods in a systematic manner, organized by their domain of application including functional connectivity, anatomical structure, and electrical-based analysis. We also outline the limitations of existing techniques and discuss potential directions for future research.

show abstract

Section: Research Challenges and Future Directionsmentioning

confidence: 99%

Graph-Based Deep Learning for Medical Diagnosis and Analysis: Past, Present and Future

Ahmedt-Aristizabal

Armin

Denman

et al. 2021

Sensors

136

View full text Add to dashboard Cite

show abstract

“…In such cases it would be ideal if we could learn these things. In electroencephalographic scans, for example, the relationship of the sensors to one another changes according to perspective [82]; but learning these structures from the data can reveal discriminative structures that help in the resolution of a task [83], [84]. Alternatively we could use such techniques to infer from a set of data over time the interactions of entities, as in physical systems [85]- [89] or multi-agent systems [90].…”

Section: Challenges In Graph Deep Learningmentioning

confidence: 99%

“…For temporal problems in physical systems this means discovering the fixed rules governing the interaction of physical objects over time, and using these rules to extrapolate to the objects' future interactions [85]- [89]. Applied to electroencephalographic (EEG) data, one might discover the discriminative structures that putatively correspond to specific psychological processes [83], [84]. Discovering graph structure in EEGs would be particularly useful in an area where several possible graph connectivities exist to represent brain signals, each capturing a different neurological perspective [82].…”

Section: Graph Estimationmentioning

confidence: 99%

See 1 more Smart Citation

Graph Deep Learning: State of the Art and Challenges

2021

View full text Add to dashboard Cite

“…Li et al [72] proposed an edge-aware ST-GCN for EEG classification, where the EEG is represented as frames of a graph. The authors selected the dataset for the task of imagining opening and closing the left or right fist [132], [133].…”

Section: B Electrical-based Analysismentioning

confidence: 99%

Graph-Based Deep Learning for Medical Diagnosis and Analysis: Past, Present and Future

Ahmedt-Aristizabal,

Armin,

Denman

et al. 2021

Preprint

View full text Add to dashboard Cite

With the advances of data-driven machine learning research, a wide variety of prediction problems have been tackled. It has become critical to explore how machine learning and specifically deep learning methods can be exploited to analyse healthcare data. A major limitation of existing methods has been the focus on grid-like data; however, the structure of physiological recordings are often irregular and unordered which makes it difficult to conceptualise them as a matrix. As such, graph neural networks have attracted significant attention by exploiting implicit information that resides in a biological system, with interactive nodes connected by edges whose weights can be either temporal associations or anatomical junctions. In this survey, we thoroughly review the different types of graph architectures and their applications in healthcare. We provide an overview of these methods in a systematic manner, organized by their domain of application including functional connectivity, anatomical structure and electrical-based analysis. We also outline the limitations of existing techniques and discuss potential directions for future research.

show abstract

Classify EEG and Reveal Latent Graph Structure with Spatio-Temporal Graph Convolutional Neural Network

Cited by 18 publications

References 15 publications

Graph-Based Deep Learning for Medical Diagnosis and Analysis: Past, Present and Future

Graph-Based Deep Learning for Medical Diagnosis and Analysis: Past, Present and Future

Graph Deep Learning: State of the Art and Challenges

Graph-Based Deep Learning for Medical Diagnosis and Analysis: Past, Present and Future

Contact Info

Product

Resources

About