EEG Source Localization Using Spatio-Temporal Neural Network
|
摘要
Source localization of focal electrical activity from scalp electroencephalogram(sEEG) signal is generally modeled as an inverse problem that is highly ill-posed. In this paper, a novel source localization method is proposed to model the EEG inverse problem using spatio-temporal long-short term memory recurrent neural networks(LSTM). The network model consists of two parts, sEEG encoding and source decoding, to model the sEEG signal and receive the regression of source location. As there does not exist enough annotated sEEG signals correspond to specific source locations, simulated data is generated with forward model using finite element method(FEM) to act as a part of training signals. A framework for source localization is proposed to estimate the source position based on simulated training data. Experiments are done on simulated testing data. The results on simulated data exhibit good robustness on noise signal, and the proposed network solves the EEG inverse problem with spatio-temporal deep network. The result show that the proposed method overcomes the highly ill-posed linear inverse problem with data driven learning.
Source localization of focal electrical activity from scalp electroencephalogram(sEEG) signal is generally modeled as an inverse problem that is highly ill-posed. In this paper, a novel source localization method is proposed to model the EEG inverse problem using spatio-temporal long-short term memory recurrent neural networks(LSTM). The network model consists of two parts, sEEG encoding and source decoding, to model the sEEG signal and receive the regression of source location. As there does not exist enough annotated sEEG signals correspond to specific source locations, simulated data is generated with forward model using finite element method(FEM) to act as a part of training signals. A framework for source localization is proposed to estimate the source position based on simulated training data. Experiments are done on simulated testing data. The results on simulated data exhibit good robustness on noise signal, and the proposed network solves the EEG inverse problem with spatio-temporal deep network. The result show that the proposed method overcomes the highly ill-posed linear inverse problem with data driven learning.
Source localization of focal electrical activity from scalp electroencephalogram(sEEG) signal is generally modeled as an inverse problem that is highly ill-posed. In this paper, a novel source localization method is proposed to model the EEG inverse problem using spatio-temporal long-short term memory recurrent neural networks(LSTM). The network model consists of two parts, sEEG encoding and source decoding, to model the sEEG signal and receive the regression of source location. As there does not exist enough annotated sEEG signals correspond to specific source locations, simulated data is generated with forward model using finite element method(FEM) to act as a part of training signals. A framework for source localization is proposed to estimate the source position based on simulated training data. Experiments are done on simulated testing data. The results on simulated data exhibit good robustness on noise signal, and the proposed network solves the EEG inverse problem with spatio-temporal deep network. The result show that the proposed method overcomes the highly ill-posed linear inverse problem with data driven learning.
引文
[1]B.He,A.Sohrabpour,E.Brown,and Z.Liu,“Electrophysiological Source Imaging:A Noninvasive Window to Brain Dynamics[J],”Annual Review of Biomedical Engineering,vol.20,no.1,2018,pp.171-196.
[2]B.He,T.Musha,Y.Okamoto,S.Homma,Y.Nakajima and T.Sato,“Electric dipole tracing in the brain by means of the boundary element method and its accuracy[J],”IEEE Transactions on Biomedical Engineering,vol.34,no.6,1987,pp.406-414.
[3]J.Mosher,P.Lewis and R.Leahy,“Multiple dipole modeling and localization from spatio-temporal MEG data[J],”IEEE Transactions on Biomedical Engineering,vol.39,no.6,1992,pp.541-557.
[4]M.H?m?l?inen and R.Ilmoniemi,“Interpreting magnetic fields of the brain:minimum norm estimates[J],”Medical&Biological Engineering&Computing,vol.32,no.1,1994,pp.35-42.
[5]A.Dale and M.Sereno,“Improved Localizadon of Cortical Activity by Combining EEG and MEGwith MRI Cortical Surface Reconstruction:ALinear Approach[J],”Journal of Cognitive Neuroscience,vol.5,no.2,1993,pp.162-176.
[6]R.Pascual-Marqui,C.Michel and D.Lehmann,“Low resolution electromagnetic tomography:a new method for localizing electrical activity in the brain[J],”International Journal of Psychophysiology,vol.18,no.1,1994,pp.49-65.
[7]R.Pascual-Marqui et al,“Standardized low-resolution brain electromagnetic tomography(sloreta):technical details[J],”Methods Find Exp Clin Pharmacol,vol.24(Suppl D),2002,pp.5-12.
[8]R.Pascual-Marqui et al.,“Assessing interactions in the brain with exact low-resolution electromagnetic tomography[J],”Philosophical Transactions of the Royal Society A:Mathematical,Physical and Engineering Sciences,vol.369,no.1952,2011,pp.3768-3784.
[9]L.Ding and B.He,“Sparse source imaging in electroencephalography with accurate field modeling[J],”Human Brain Mapping,vol.29,no.9,2008,pp.1053-1067.
[10]M.SEEGER,“GAUSSIAN PROCESSES FOR MA-CHINE LEARNING[J],”International Journal of Neural Systems,vol.14,no.02,2004,pp.69-106.
[11]D.Wipf and S.Nagarajan,“A unified Bayesian framework for MEG/EEG source imaging[J],”NeuroImage,vol.44,no.3,2009,pp.947-966.
[12]S.Baillet and L.Garnero,“A Bayesian approach to introducing anatomo-functional priors in the EEG/MEG inverse problem[J],”IEEE Transactions on Biomedical Engineering,vol.44,no.5,1997,pp.374-385.
[13]J.Daunizeau,J.Mattout,D.Clonda,B.Goulard,H.Benali and J.Lina,“Bayesian Spatio-Temporal Approach for EEG Source Reconstruction:Conciliating ECD and Distributed Models[J],”IEEETransactions on Biomedical Engineering,vol.53,no.3,2006,pp.503-516.
[14]C.Lamus,M.H?m?l?inen,S.Temereanca,E.Brown and P.Purdon,“A spatiotemporal dynamic distributed solution to the MEG inverse problem[J],”NeuroImage,vol.63,no.2,2012,pp.894-909.
[15]K.Liu,Z.Yu,W.Wu,Z.Gu and Y.Li,“STRAPS:AFully Data-Driven Spatio-Temporally Regularized Algorithm for M/EEG Patch Source Imaging[J],”International Journal of Neural Systems,vol.25,no.04,2015,pp.1550016.
[16]A.Gramfort,D.Strohmeier,J.Haueisen,M.H?m?l?inen and M.Kowalski,“Time-frequency mixed-norm estimates:Sparse M/EEG imaging with non-stationary source activations[J],”Neuro Image,vol.70,2013,pp.410-422.
[17]E.Giraldo-Suarez,J.Martinez-Vargas and G.Castellanos-Dominguez,“Reconstruction of Neural Activity from EEG Data Using Dynamic Spatiotemporal Constraints[J],”International Journal of Neural Systems,vol.26,no.07,2016,pp.1650026.
[18]K.Liu,Z.Yu,W.Wu,Z.Gu,Y.Li and S.Nagarajan,“Bayesian electromagnetic spatio-temporal imaging of extended sources with Markov Random Field and temporal basis expansion[J],”NeuroImage,vol.139,2016,pp.385-404.
[19]D.Pa z-L i n a r e s,M.Ve g a-H e r nán d e z,P.R o j a s-Lóp e z,P.Va l dés-H e r nán d e z,E.Martínez-Montes and P.Valdés-Sosa,“Spatio Temporal EEG Source Imaging with the Hierarchical Bayesian Elastic Net and Elitist Lasso Models[J],”Frontiers in Neuroscience,vol.11,2017.
[20]Feng.Liu,et al.“Task-related eeg source localization via graph regularized low-rank representation model[J],”bioRxiv(2018):246579.
[21]Y.Kinouchi,G.Ohara,H.Nagashino,T.Soga,F.Shichijo and K.Matsumoto,“Dipole source localization of MEG by BP neural networks[J],”Brain Topography,vol.8,no.3,1996,pp.317-321.
[22]G.Hoey et al.,“EEG dipole source localization using artificial neural networks[J],”Physics in Medicine and Biology,vol.45,no.4,2000,pp.997-1011.
[23]S.Jun,B.Pearlmutter and G.Nolte,“Fast accurate MEG source localization using a multilayer perceptron trained with real brain noise[J],”Physics in Medicine and Biology,vol.47,no.14,2002,pp.2547-2560.
[24]S.Jun and B.Pearlmutter,“Fast robust subject-independent magnetoencephalographic source localization using an artificial neural network[J],”Human Brain Mapping,vol.24,no.1,2004,pp.21-34.
[25]J.H.Rick Chang,Chun-Liang Li,Barnabas Poczos,and B.V.K.Vijaya Kumar,“One Network to Solve Them All--Solving Linear Inverse Problems Using Deep Projection Models[C],”2017IEEE International Conference on Computer Vision(ICCV),IEEE,2017,pp.5888-5897.
[26]K.Jin,M.McCann,E.Froustey and M.Unser,“Deep Convolutional Neural Network for Inverse Problems in Imaging[J],”IEEE Transactions on Image Processing,vol.26,no.9,2017,pp.4509-4522.
[27]J.Adler and O.?ktem,“Solving ill-posed inverse problems using iterative deep neural networks[J],”Inverse Problems,vol.33,no.12,2017,p.124007.
[28]A.Lucas,M.Iliadis,R.Molina and A.Katsaggelos,“Using Deep Neural Networks for Inverse Problems in Imaging:Beyond Analytical Methods[J],”IEEE Signal Processing Magazine,vol.35,no.1,2018,pp.20-36.
[29]T.Yang,L.Yu,Q.Jin,L.Wu and B.He,“Localization of Origins of Premature Ventricular Contraction by Means of Convolutional Neural Network From 12-Lead ECG[J],”IEEE Transactions on Biomedical Engineering,vol.65,no.7,2018,pp.1662-1671.
[30]A.Graves,A.Mohamed,&G.Hinton,“Speech recognition with deep recurrent neural networks[C],”2013 IEEE international conference on acoustics,speech and signal processing(ICASSP),IEEE,2013,pp.6645-6649.
[31]T.Mikolov,M.Karafiát,L.Burget,J.?ernocky,&S.Khudanpur,"Recurrent neural network based language model[C],"Eleventh annual conference of the international speech communication association,2010.
[32]S.Hochreiter and J.Schmidhuber,“Long ShortTerm Memory[J],”Neural Computation,vol.9,no.8,1997,pp.1735-1780.
[33]I.Sutskever,O.Vinyals,and Q.V.Le,“Sequence to sequence learning with neural networks[C],”Advances in neural information processing systems,2014,pp.3104-3112.
[34]W.Zaremba,and I.Sutskever."Learning to execute[J],"arXiv preprint arXiv:1410.4615,2014.
[35]R.Oostenveld,P.Fries,E.Maris and J.Schoffelen,“FieldTrip:Open Source Software for Advanced Analysis of MEG,EEG,and Invasive Electrophysiological Data[J],”Computational Intelligence and Neuroscience,vol.2011,2011,pp.1-9.
[36]S.Gabriel,R.Lau and C.Gabriel,“The dielectric properties of biological tissues:II.Measurements in the frequency range 10 Hz to 20GHz[J],”Physics in Medicine and Biology,vol.41,no.11,1996,pp.2251-2269.
[37]L.Manola,B.Roelofsen,J.Holsheimer,E.Marani and J.Geelen,“Modelling motor cortex stimulation for chronic pain control:Electrical potential field,activating functions and responses of simple nerve fibre models[J],”Medical&Biological Engineering&Computing,vol.43,no.3,2005,pp.335-343.
[38]K.Simonyan,and A.Zisserman.“Very deep convolutional networks for large-scale image recognition[J],”arXiv preprint arXiv:1409.1556,2014.
[2]B.He,T.Musha,Y.Okamoto,S.Homma,Y.Nakajima and T.Sato,“Electric dipole tracing in the brain by means of the boundary element method and its accuracy[J],”IEEE Transactions on Biomedical Engineering,vol.34,no.6,1987,pp.406-414.
[3]J.Mosher,P.Lewis and R.Leahy,“Multiple dipole modeling and localization from spatio-temporal MEG data[J],”IEEE Transactions on Biomedical Engineering,vol.39,no.6,1992,pp.541-557.
[4]M.H?m?l?inen and R.Ilmoniemi,“Interpreting magnetic fields of the brain:minimum norm estimates[J],”Medical&Biological Engineering&Computing,vol.32,no.1,1994,pp.35-42.
[5]A.Dale and M.Sereno,“Improved Localizadon of Cortical Activity by Combining EEG and MEGwith MRI Cortical Surface Reconstruction:ALinear Approach[J],”Journal of Cognitive Neuroscience,vol.5,no.2,1993,pp.162-176.
[6]R.Pascual-Marqui,C.Michel and D.Lehmann,“Low resolution electromagnetic tomography:a new method for localizing electrical activity in the brain[J],”International Journal of Psychophysiology,vol.18,no.1,1994,pp.49-65.
[7]R.Pascual-Marqui et al,“Standardized low-resolution brain electromagnetic tomography(sloreta):technical details[J],”Methods Find Exp Clin Pharmacol,vol.24(Suppl D),2002,pp.5-12.
[8]R.Pascual-Marqui et al.,“Assessing interactions in the brain with exact low-resolution electromagnetic tomography[J],”Philosophical Transactions of the Royal Society A:Mathematical,Physical and Engineering Sciences,vol.369,no.1952,2011,pp.3768-3784.
[9]L.Ding and B.He,“Sparse source imaging in electroencephalography with accurate field modeling[J],”Human Brain Mapping,vol.29,no.9,2008,pp.1053-1067.
[10]M.SEEGER,“GAUSSIAN PROCESSES FOR MA-CHINE LEARNING[J],”International Journal of Neural Systems,vol.14,no.02,2004,pp.69-106.
[11]D.Wipf and S.Nagarajan,“A unified Bayesian framework for MEG/EEG source imaging[J],”NeuroImage,vol.44,no.3,2009,pp.947-966.
[12]S.Baillet and L.Garnero,“A Bayesian approach to introducing anatomo-functional priors in the EEG/MEG inverse problem[J],”IEEE Transactions on Biomedical Engineering,vol.44,no.5,1997,pp.374-385.
[13]J.Daunizeau,J.Mattout,D.Clonda,B.Goulard,H.Benali and J.Lina,“Bayesian Spatio-Temporal Approach for EEG Source Reconstruction:Conciliating ECD and Distributed Models[J],”IEEETransactions on Biomedical Engineering,vol.53,no.3,2006,pp.503-516.
[14]C.Lamus,M.H?m?l?inen,S.Temereanca,E.Brown and P.Purdon,“A spatiotemporal dynamic distributed solution to the MEG inverse problem[J],”NeuroImage,vol.63,no.2,2012,pp.894-909.
[15]K.Liu,Z.Yu,W.Wu,Z.Gu and Y.Li,“STRAPS:AFully Data-Driven Spatio-Temporally Regularized Algorithm for M/EEG Patch Source Imaging[J],”International Journal of Neural Systems,vol.25,no.04,2015,pp.1550016.
[16]A.Gramfort,D.Strohmeier,J.Haueisen,M.H?m?l?inen and M.Kowalski,“Time-frequency mixed-norm estimates:Sparse M/EEG imaging with non-stationary source activations[J],”Neuro Image,vol.70,2013,pp.410-422.
[17]E.Giraldo-Suarez,J.Martinez-Vargas and G.Castellanos-Dominguez,“Reconstruction of Neural Activity from EEG Data Using Dynamic Spatiotemporal Constraints[J],”International Journal of Neural Systems,vol.26,no.07,2016,pp.1650026.
[18]K.Liu,Z.Yu,W.Wu,Z.Gu,Y.Li and S.Nagarajan,“Bayesian electromagnetic spatio-temporal imaging of extended sources with Markov Random Field and temporal basis expansion[J],”NeuroImage,vol.139,2016,pp.385-404.
[19]D.Pa z-L i n a r e s,M.Ve g a-H e r nán d e z,P.R o j a s-Lóp e z,P.Va l dés-H e r nán d e z,E.Martínez-Montes and P.Valdés-Sosa,“Spatio Temporal EEG Source Imaging with the Hierarchical Bayesian Elastic Net and Elitist Lasso Models[J],”Frontiers in Neuroscience,vol.11,2017.
[20]Feng.Liu,et al.“Task-related eeg source localization via graph regularized low-rank representation model[J],”bioRxiv(2018):246579.
[21]Y.Kinouchi,G.Ohara,H.Nagashino,T.Soga,F.Shichijo and K.Matsumoto,“Dipole source localization of MEG by BP neural networks[J],”Brain Topography,vol.8,no.3,1996,pp.317-321.
[22]G.Hoey et al.,“EEG dipole source localization using artificial neural networks[J],”Physics in Medicine and Biology,vol.45,no.4,2000,pp.997-1011.
[23]S.Jun,B.Pearlmutter and G.Nolte,“Fast accurate MEG source localization using a multilayer perceptron trained with real brain noise[J],”Physics in Medicine and Biology,vol.47,no.14,2002,pp.2547-2560.
[24]S.Jun and B.Pearlmutter,“Fast robust subject-independent magnetoencephalographic source localization using an artificial neural network[J],”Human Brain Mapping,vol.24,no.1,2004,pp.21-34.
[25]J.H.Rick Chang,Chun-Liang Li,Barnabas Poczos,and B.V.K.Vijaya Kumar,“One Network to Solve Them All--Solving Linear Inverse Problems Using Deep Projection Models[C],”2017IEEE International Conference on Computer Vision(ICCV),IEEE,2017,pp.5888-5897.
[26]K.Jin,M.McCann,E.Froustey and M.Unser,“Deep Convolutional Neural Network for Inverse Problems in Imaging[J],”IEEE Transactions on Image Processing,vol.26,no.9,2017,pp.4509-4522.
[27]J.Adler and O.?ktem,“Solving ill-posed inverse problems using iterative deep neural networks[J],”Inverse Problems,vol.33,no.12,2017,p.124007.
[28]A.Lucas,M.Iliadis,R.Molina and A.Katsaggelos,“Using Deep Neural Networks for Inverse Problems in Imaging:Beyond Analytical Methods[J],”IEEE Signal Processing Magazine,vol.35,no.1,2018,pp.20-36.
[29]T.Yang,L.Yu,Q.Jin,L.Wu and B.He,“Localization of Origins of Premature Ventricular Contraction by Means of Convolutional Neural Network From 12-Lead ECG[J],”IEEE Transactions on Biomedical Engineering,vol.65,no.7,2018,pp.1662-1671.
[30]A.Graves,A.Mohamed,&G.Hinton,“Speech recognition with deep recurrent neural networks[C],”2013 IEEE international conference on acoustics,speech and signal processing(ICASSP),IEEE,2013,pp.6645-6649.
[31]T.Mikolov,M.Karafiát,L.Burget,J.?ernocky,&S.Khudanpur,"Recurrent neural network based language model[C],"Eleventh annual conference of the international speech communication association,2010.
[32]S.Hochreiter and J.Schmidhuber,“Long ShortTerm Memory[J],”Neural Computation,vol.9,no.8,1997,pp.1735-1780.
[33]I.Sutskever,O.Vinyals,and Q.V.Le,“Sequence to sequence learning with neural networks[C],”Advances in neural information processing systems,2014,pp.3104-3112.
[34]W.Zaremba,and I.Sutskever."Learning to execute[J],"arXiv preprint arXiv:1410.4615,2014.
[35]R.Oostenveld,P.Fries,E.Maris and J.Schoffelen,“FieldTrip:Open Source Software for Advanced Analysis of MEG,EEG,and Invasive Electrophysiological Data[J],”Computational Intelligence and Neuroscience,vol.2011,2011,pp.1-9.
[36]S.Gabriel,R.Lau and C.Gabriel,“The dielectric properties of biological tissues:II.Measurements in the frequency range 10 Hz to 20GHz[J],”Physics in Medicine and Biology,vol.41,no.11,1996,pp.2251-2269.
[37]L.Manola,B.Roelofsen,J.Holsheimer,E.Marani and J.Geelen,“Modelling motor cortex stimulation for chronic pain control:Electrical potential field,activating functions and responses of simple nerve fibre models[J],”Medical&Biological Engineering&Computing,vol.43,no.3,2005,pp.335-343.
[38]K.Simonyan,and A.Zisserman.“Very deep convolutional networks for large-scale image recognition[J],”arXiv preprint arXiv:1409.1556,2014.