Multiway analysis of epilepsy tensors

By Evrim Acar, Canan Aykut-Bingöl, H. Bingol, Rasmus Bro, Bülent Yener. ISMB/ECCB, 2007

298 citations recorded by the source corpus.

MOTIVATION The success or failure of an epilepsy surgery depends greatly on the localization of epileptic focus (origin of a seizure). We address the problem of identification of a seizure origin through an analysis of ictal electroencephalogram (EEG), which is proven to be an effective standard in epileptic focus localization. SUMMARY With a goal of developing an automated and robust way of visual analysis of large amounts of EEG data, we propose a novel approach based on multiway models to study epilepsy seizure structure. Our contributions are 3-fold. First, we construct an Epilepsy Tensor with three modes, i.e. time samples, scales and electrodes, through wavelet analysis of multi-channel ictal EEG. Second, we demonstrate that multiway analysis techniques, in particular parallel factor analysis (PARAFAC), provide promising results in modeling the complex structure of an epilepsy seizure, localizing a seizure origin and extracting artifacts. Third, we introduce an approach for removing artifacts using multilinear subspace analysis and discuss its merits and drawbacks. RESULTS Ictal EEG analysis of 10 seizures from 7 patients are included in this study. Our results for 8 seizures match with clinical observations in terms of seizure origin and extracted artifacts. On the other hand, for 2 of the seizures, seizure localization is not achieved using an initial trial of PARAFAC modeling. In these cases, first, we apply an artifact removal method and subsequently apply the PARAFAC model on the epilepsy tensor from which potential artifacts have been removed. This method successfully identifies the seizure origin in both cases.

Subfields: Cognitive Neuroscience, Psychiatry and Mental health

View Paper: https://www.semanticscholar.org/paper/cafd21a9559255e496465ece76d398f51cbd3d39

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cognitive Neuroscience

In Psychiatry and Mental health

Citations in the corpus

Focal Onset Seizure Prediction Using Convolutional Networks (2018, methodology): “This was motivated by previous success with the use of tensor decomposition and analysis on wavelet tensors for the seizure detection problem [40] and with a similar approach to genomic data [1].”
Best Low Multilinear Rank Approximation of Higher-Order Tensors, Based on the Riemannian Trust-Region Scheme (2011, background): “Fields of application of higher-order tensors also include chemometrics [42], biomedical signal processing [31, 15, 16, 4], telecommunications [40, 41, 10], scientific computing [7, 6, 20, 35], and image processing [48].”
Limestone: High-throughput candidate phenotype generation via tensor factorization (2014, methodology): “The CP decomposition has been used to complete missing data in medical questionnaires [56], localize and extract artifacts from EEG data to analyze epileptic seizures [57,58], and as an exploratory decomposition tool for wavelet-transformed multi-channel EEG data [59].”
Scalable Tensor Factorizations for Incomplete Data (2010, background): “We focus here on the CANDECOMP/PARAFAC (CP) tensor decomposition [14, 15], which is a tensor model commonly used in various applications [6, 16, 17, 18, 19].”
Differential-geometric Newton method for the best rank-(R1, R2, R3) approximation of tensors (2008, background): “PARAFAC can also be used for epileptic seizure onset localisation [3, 20, 21], since only one of its components is related to the seizure activity.”
Tensors for Data Mining and Data Fusion (2016, methodology): “Furthermore, we can assume that U i are orthogonal and matrices, and as stated in Acar et al. [2007], they reﬂect the main subspace variation in each mode assuming a multilinear structure.”
Unsupervised Discovery of Demixed, Low-Dimensional Neural Dynamics across Multiple Timescales through Tensor Component Analysis (2017, background): “Several studies have applied tensor decompositions to EEG and fMRI data, most typically to model differences across subjects or Fourier/wavelet transformed signals (Mørup et al., 2006; Acar et al., 2007; Cong et al., 2015; Hunyadi et al., 2017), rather than across trials (Andersen and Rayens, 2004;”
Multimodal Data Fusion: An Overview of Methods, Challenges, and Prospects (2015, background): “However, it has been proposed to augment this data using a third type of diversity, electrode ⇥ time ⇥ frequency [101]– [103].”
Unsupervised Multiway Data Analysis: A Literature Survey (2009, methodology): “Core consistency diagnostic has been commonly applied in the literature [3], [5], [20], [21].”
Tensor Decompositions and Applications (2009, background): “Examples include signal processing [62, 196, 47, 43, 68, 80, 173, 60, 61], numerical linear algebra [87, 63, 64, 132, 244, 133, 149], computer vision [229, 230, 231, 190, 236, 237, 193, 102, 235], numerical analysis [22, 108, 23, 89, 114, 88, 115], data mining [190, 4, 157, 211, 5, 209, 210, 44, 14]”
Tensor decomposition of EEG signals: a brief review. (2015, unknown): “Such a tensor has been decomposed using the CP model (Acar et al., 2007a,b; De Vos et al., 2007b; Deburchgraeve et al., 2009; Miwakeichi et al., 2004; Wang et al., 2012).”
On the best low multilinear rank approximation of higher-order tensors (2010, methodology): “The best low multilinear rank approximation of tensors is used for signal subspace estimation [60, 61, 52, 67, 51, 35] and as a dimensionality reduction tool for tensors with high dimensions [27, 4 , 29, 30, 52, 67, 51], including simultaneous dimensionality reduction of a matrix and a tensor [27].”
A scalable optimization approach for fitting canonical tensor decompositions (2011, methodology): “As an example of using the core consistency diagnostic, see [1] for further discussion in the context of extracting brain activities through the analysis of EEG (electroencephalography) signals without any prior information about the number of activities.”