Temporal Graph Convolutional Networks for Automatic Seizure Detection

Ian C. Covert, Balu Krishnan, Imad Najm, Jiening Zhan, Matthew Shore, John Hixson, Ming Jack Po
Proceedings of the 4th Machine Learning for Healthcare Conference, PMLR 106:160-180, 2019.

Abstract

Seizure detection from EEGs is a challenging and time-consuming clinical problem that would benefit from the development of automated algorithms. EEGs can be viewed as structural time series, because they are multivariate time series where the placement of leads on a patient’s scalp provides prior information about the structure of interactions. Commonly used deep learning models for time series do not offer a way to leverage structural information, but this would be desirable in a model for structural time series. To address this challenge, we propose the temporal graph convolutional network (TGCN), a model that leverages temporal and structural information and has relatively few parameters. TGCN applies feature extraction operations that are localized and shared over both time and space, thereby providing a useful inductive bias in tasks where similar features are expected to be discriminative across the different sequences. In our experiments we focus on metrics that are most important to seizure detection, and demonstrate that TGCN matches the performance of related models that have been shown to be state-of-the-art in other tasks. Additionally, we investigate interpretability advantages of TGCN by exploring approaches for helping clinicians determine when precisely seizures occur, and the parts of the brain that are most involved.

Cite this Paper

BibTeX
@InProceedings{pmlr-v106-covert19a, title = {Temporal Graph Convolutional Networks for Automatic Seizure Detection}, author = {Covert, Ian C. and Krishnan, Balu and Najm, Imad and Zhan, Jiening and Shore, Matthew and Hixson, John and Po, Ming Jack}, booktitle = {Proceedings of the 4th Machine Learning for Healthcare Conference}, pages = {160--180}, year = {2019}, editor = {Doshi-Velez, Finale and Fackler, Jim and Jung, Ken and Kale, David and Ranganath, Rajesh and Wallace, Byron and Wiens, Jenna}, volume = {106}, series = {Proceedings of Machine Learning Research}, month = {09--10 Aug}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v106/covert19a/covert19a.pdf}, url = {https://proceedings.mlr.press/v106/covert19a.html}, abstract = {Seizure detection from EEGs is a challenging and time-consuming clinical problem that would benefit from the development of automated algorithms. EEGs can be viewed as structural time series, because they are multivariate time series where the placement of leads on a patient’s scalp provides prior information about the structure of interactions. Commonly used deep learning models for time series do not offer a way to leverage structural information, but this would be desirable in a model for structural time series. To address this challenge, we propose the temporal graph convolutional network (TGCN), a model that leverages temporal and structural information and has relatively few parameters. TGCN applies feature extraction operations that are localized and shared over both time and space, thereby providing a useful inductive bias in tasks where similar features are expected to be discriminative across the different sequences. In our experiments we focus on metrics that are most important to seizure detection, and demonstrate that TGCN matches the performance of related models that have been shown to be state-of-the-art in other tasks. Additionally, we investigate interpretability advantages of TGCN by exploring approaches for helping clinicians determine when precisely seizures occur, and the parts of the brain that are most involved.} }
Endnote
%0 Conference Paper %T Temporal Graph Convolutional Networks for Automatic Seizure Detection %A Ian C. Covert %A Balu Krishnan %A Imad Najm %A Jiening Zhan %A Matthew Shore %A John Hixson %A Ming Jack Po %B Proceedings of the 4th Machine Learning for Healthcare Conference %C Proceedings of Machine Learning Research %D 2019 %E Finale Doshi-Velez %E Jim Fackler %E Ken Jung %E David Kale %E Rajesh Ranganath %E Byron Wallace %E Jenna Wiens %F pmlr-v106-covert19a %I PMLR %P 160--180 %U https://proceedings.mlr.press/v106/covert19a.html %V 106 %X Seizure detection from EEGs is a challenging and time-consuming clinical problem that would benefit from the development of automated algorithms. EEGs can be viewed as structural time series, because they are multivariate time series where the placement of leads on a patient’s scalp provides prior information about the structure of interactions. Commonly used deep learning models for time series do not offer a way to leverage structural information, but this would be desirable in a model for structural time series. To address this challenge, we propose the temporal graph convolutional network (TGCN), a model that leverages temporal and structural information and has relatively few parameters. TGCN applies feature extraction operations that are localized and shared over both time and space, thereby providing a useful inductive bias in tasks where similar features are expected to be discriminative across the different sequences. In our experiments we focus on metrics that are most important to seizure detection, and demonstrate that TGCN matches the performance of related models that have been shown to be state-of-the-art in other tasks. Additionally, we investigate interpretability advantages of TGCN by exploring approaches for helping clinicians determine when precisely seizures occur, and the parts of the brain that are most involved.
APA
Covert, I.C., Krishnan, B., Najm, I., Zhan, J., Shore, M., Hixson, J. & Po, M.J.. (2019). Temporal Graph Convolutional Networks for Automatic Seizure Detection. Proceedings of the 4th Machine Learning for Healthcare Conference, in Proceedings of Machine Learning Research 106:160-180 Available from https://proceedings.mlr.press/v106/covert19a.html.

Related Material