Unsupervised representation learning of spontaneous MEG data with nonlinear ICA
Resting-state magnetoencephalography (MEG) data show complex but structured spatiotemporal patterns. However, the neurophysiological basis of these signal patterns is not fully known and the underlying signal sources are mixed in MEG measurements. Here, we developed a method based on the nonlinear i...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2023-07-01
|
| Series: | NeuroImage |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1053811923002938 |
| _version_ | 1827947415787798528 |
|---|---|
| author | Yongjie Zhu Tiina Parviainen Erkka Heinilä Lauri Parkkonen Aapo Hyvärinen |
| author_facet | Yongjie Zhu Tiina Parviainen Erkka Heinilä Lauri Parkkonen Aapo Hyvärinen |
| author_sort | Yongjie Zhu |
| collection | DOAJ |
| description | Resting-state magnetoencephalography (MEG) data show complex but structured spatiotemporal patterns. However, the neurophysiological basis of these signal patterns is not fully known and the underlying signal sources are mixed in MEG measurements. Here, we developed a method based on the nonlinear independent component analysis (ICA), a generative model trainable with unsupervised learning, to learn representations from resting-state MEG data. After being trained with a large dataset from the Cam-CAN repository, the model has learned to represent and generate patterns of spontaneous cortical activity using latent nonlinear components, which reflects principal cortical patterns with specific spectral modes. When applied to the downstream classification task of audio-visual MEG, the nonlinear ICA model achieves competitive performance with deep neural networks despite limited access to labels. We further validate the generalizability of the model across different datasets by applying it to an independent neurofeedback dataset for decoding the subject's attentional states, providing a real-time feature extraction and decoding mindfulness and thought-inducing tasks with an accuracy of around 70% at the individual level, which is much higher than obtained by linear ICA or other baseline methods. Our results demonstrate that nonlinear ICA is a valuable addition to existing tools, particularly suited for unsupervised representation learning of spontaneous MEG activity which can then be applied to specific goals or tasks when labelled data are scarce. |
| first_indexed | 2024-04-09T12:39:39Z |
| format | Article |
| id | doaj.art-52666a3240f14ff6a85a187b0297c468 |
| institution | Directory Open Access Journal |
| issn | 1095-9572 |
| language | English |
| last_indexed | 2024-04-09T12:39:39Z |
| publishDate | 2023-07-01 |
| publisher | Elsevier |
| record_format | Article |
| series | NeuroImage |
| spelling | doaj.art-52666a3240f14ff6a85a187b0297c4682023-05-15T04:13:56ZengElsevierNeuroImage1095-95722023-07-01274120142Unsupervised representation learning of spontaneous MEG data with nonlinear ICAYongjie Zhu0Tiina Parviainen1Erkka Heinilä2Lauri Parkkonen3Aapo Hyvärinen4Department of Computer Science, University of Helsinki, 00560 Helsinki, Finland; Department of Neuroscience and Biomedical Engineering, Aalto University, 00076 Espoo, Finland; Corresponding author.Centre for Interdisciplinary Brain Research, Department of Psychology, University of Jyväskylä, 40014 Jyväskylä, FinlandCentre for Interdisciplinary Brain Research, Department of Psychology, University of Jyväskylä, 40014 Jyväskylä, FinlandDepartment of Neuroscience and Biomedical Engineering, Aalto University, 00076 Espoo, FinlandDepartment of Computer Science, University of Helsinki, 00560 Helsinki, FinlandResting-state magnetoencephalography (MEG) data show complex but structured spatiotemporal patterns. However, the neurophysiological basis of these signal patterns is not fully known and the underlying signal sources are mixed in MEG measurements. Here, we developed a method based on the nonlinear independent component analysis (ICA), a generative model trainable with unsupervised learning, to learn representations from resting-state MEG data. After being trained with a large dataset from the Cam-CAN repository, the model has learned to represent and generate patterns of spontaneous cortical activity using latent nonlinear components, which reflects principal cortical patterns with specific spectral modes. When applied to the downstream classification task of audio-visual MEG, the nonlinear ICA model achieves competitive performance with deep neural networks despite limited access to labels. We further validate the generalizability of the model across different datasets by applying it to an independent neurofeedback dataset for decoding the subject's attentional states, providing a real-time feature extraction and decoding mindfulness and thought-inducing tasks with an accuracy of around 70% at the individual level, which is much higher than obtained by linear ICA or other baseline methods. Our results demonstrate that nonlinear ICA is a valuable addition to existing tools, particularly suited for unsupervised representation learning of spontaneous MEG activity which can then be applied to specific goals or tasks when labelled data are scarce.http://www.sciencedirect.com/science/article/pii/S1053811923002938Nonlinear independent component analysis (ICA)Unsupervised learningDeep generative modelResting-state networkNon-stationarityNeurofeedback |
| spellingShingle | Yongjie Zhu Tiina Parviainen Erkka Heinilä Lauri Parkkonen Aapo Hyvärinen Unsupervised representation learning of spontaneous MEG data with nonlinear ICA NeuroImage Nonlinear independent component analysis (ICA) Unsupervised learning Deep generative model Resting-state network Non-stationarity Neurofeedback |
| title | Unsupervised representation learning of spontaneous MEG data with nonlinear ICA |
| title_full | Unsupervised representation learning of spontaneous MEG data with nonlinear ICA |
| title_fullStr | Unsupervised representation learning of spontaneous MEG data with nonlinear ICA |
| title_full_unstemmed | Unsupervised representation learning of spontaneous MEG data with nonlinear ICA |
| title_short | Unsupervised representation learning of spontaneous MEG data with nonlinear ICA |
| title_sort | unsupervised representation learning of spontaneous meg data with nonlinear ica |
| topic | Nonlinear independent component analysis (ICA) Unsupervised learning Deep generative model Resting-state network Non-stationarity Neurofeedback |
| url | http://www.sciencedirect.com/science/article/pii/S1053811923002938 |
| work_keys_str_mv | AT yongjiezhu unsupervisedrepresentationlearningofspontaneousmegdatawithnonlinearica AT tiinaparviainen unsupervisedrepresentationlearningofspontaneousmegdatawithnonlinearica AT erkkaheinila unsupervisedrepresentationlearningofspontaneousmegdatawithnonlinearica AT lauriparkkonen unsupervisedrepresentationlearningofspontaneousmegdatawithnonlinearica AT aapohyvarinen unsupervisedrepresentationlearningofspontaneousmegdatawithnonlinearica |