{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/deep-learning-with-convolutional-neural-1","title":"Deep learning with convolutional neural networks for decoding and visualization of EEG pathology","arxiv_id":"1708.08012","date":"2017-08-26","proceeding":null,"authors":["Robin Tibor Schirrmeister","Lukas Gemein","Katharina Eggensperger","Frank Hutter","Tonio Ball"],"abstract":"We apply convolutional neural networks (ConvNets) to the task of\ndistinguishing pathological from normal EEG recordings in the Temple University\nHospital EEG Abnormal Corpus. We use two basic, shallow and deep ConvNet\narchitectures recently shown to decode task-related information from EEG at\nleast as well as established algorithms designed for this purpose. In decoding\nEEG pathology, both ConvNets reached substantially better accuracies (about 6%\nbetter, ~85% vs. ~79%) than the only published result for this dataset, and\nwere still better when using only 1 minute of each recording for training and\nonly six seconds of each recording for testing. We used automated methods to\noptimize architectural hyperparameters and found intriguingly different ConvNet\narchitectures, e.g., with max pooling as the only nonlinearity. Visualizations\nof the ConvNet decoding behavior showed that they used spectral power changes\nin the delta (0-4 Hz) and theta (4-8 Hz) frequency range, possibly alongside\nother features, consistent with expectations derived from spectral analysis of\nthe EEG data and from the textual medical reports. Analysis of the textual\nmedical reports also highlighted the potential for accuracy increases by\nintegrating contextual information, such as the age of subjects. In summary,\nthe ConvNets and visualization techniques used in this study constitute a next\nstep towards clinically useful automated EEG diagnosis and establish a new\nbaseline for future work on this topic.","url_abs":"http://arxiv.org/abs/1708.08012v3","url_pdf":"http://arxiv.org/pdf/1708.08012v3.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"deep-learning-with-convolutional-neural-1","repo_url":"https://github.com/robintibor/auto-eeg-diagnosis-example","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"deep-learning-with-convolutional-neural-1","repo_url":"https://github.com/DWonGH/autotuab","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"eeg-1","task_name":"EEG"},{"task_slug":"eeg","task_name":"Electroencephalogram (EEG)"}],"methods":[{"method_slug":"max-pooling","method_name":"Max Pooling"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1708.08012","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}