Academic Journal
Προβολή σε EDS

Deep Learning for Ear-EEG-Based Brain-Computer Interface: A Systematic Comparison and Design Insights.

Λεπτομέρειες βιβλιογραφικής εγγραφής
Τίτλος: Deep Learning for Ear-EEG-Based Brain-Computer Interface: A Systematic Comparison and Design Insights.
Συγγραφείς: Kim JS; Department of Computer and Information Science, Korea University, Sejong 30019, Republic of Korea., Choi SI; Medical Metrology Group, Division of Biomedical Metrology, Korea Research Institute of Standards and Science (KRISS), Daejeon 34113, Republic of Korea., Hwang HJ; Department of Electronics and Information Engineering, Korea University, Sejong 30019, Republic of Korea.; Digital Healthcare Center, Sejong Institute for Business and Technology, Korea University, Sejong 30019, Republic of Korea.; Interdisciplinary Graduate Program for Artificial Intelligence Smart Convergence Technology, Korea University, Sejong 30019, Republic of Korea., Han CH; Digital Healthcare Center, Sejong Institute for Business and Technology, Korea University, Sejong 30019, Republic of Korea.; Department of Computer Science and Software Engineering, Korea University, Sejong 30019, Republic of Korea.
Πηγή: Biosensors [Biosensors (Basel)] 2026 Aug 12; Vol. 16 (8). Date of Electronic Publication: 2026 Aug 12.
Τύπος έκδοσης: Journal Article
Γλώσσα: English
Στοιχεία περιοδικού: Publisher: MDPI Pub Country of Publication: Switzerland NLM ID: 101609191 Publication Model: Electronic Cited Medium: Internet ISSN: 2079-6374 (Electronic) Linking ISSN: 20796374 NLM ISO Abbreviation: Biosensors (Basel) Subsets: MEDLINE
Imprint Name(s): Original Publication: Basel, Switzerland : MDPI Pub.
Ιατρικοί όροι (MeSH): Brain-Computer Interfaces* , Deep Learning* , Electroencephalography* , Ear*, Humans ; Neural Networks, Computer ; Signal-To-Noise Ratio
Περίληψη: Electroencephalography (EEG) measured inside or around ears, called ear-EEG, provides a practical measurement modality for daily brain-computer interface (BCI) applications. However, reliable decoding of mental imagery remains challenging due to the limited number of channels, low signal-to-noise ratio (SNR), and substantial inter- and intra-subject variability inherent to ear-EEG. Addressing these constraints requires advanced decoding strategies specifically optimized for this signal domain. In this study, we retrospectively analyze the ear-EEG dataset of a previous study in which a real-time endogenous BCI was evaluated using conventional machine learning. Specifically, we present an offline benchmark of 23 deep neural network architectures originally developed for scalp-EEG, which were adapted to ear-EEG and evaluated under an identical validation framework. To the best of our knowledge, this is the first systematic comparison of this breadth for ear-EEG-based mental-task classification. Beyond conventional performance comparison, we identify the optimal architecture by jointly considering statistical significance and a performance-cost trade-off, incorporating classification accuracy, parameter count, and measured computational cost. Our results demonstrate that FBLightConvNet achieves the highest classification accuracy among all evaluated models and outperforms common spatial pattern-linear discriminant analysis (CSP-LDA), a widely adopted and robust conventional baseline, on all three recording days, with the difference reaching statistical significance on Days 2 and 3. Notably, many state-of-the-art scalp-EEG models fail to generalize effectively to ear-EEG, highlighting the importance of architecture selection in this domain. These findings identify the best-performing architecture in this setting and indicate which architectural characteristics support effective ear-EEG decoding. Ultimately, this study offers practical design insights and a reproducible benchmarking framework for developing lightweight and high-performance deep learning models, which we hope will support future efforts toward real-world ear-EEG-based BCI systems.
References: IEEE Rev Biomed Eng. 2009;2:187-199. (PMID: 20442804)
Front Comput Neurosci. 2023 Apr 26;17:1142948. (PMID: 37180880)
J Neural Eng. 2018 Oct;15(5):056013. (PMID: 29932424)
Hum Brain Mapp. 2017 Nov;38(11):5391-5420. (PMID: 28782865)
Neurology. 2023 Apr 25;100(17):e1750-e1762. (PMID: 36878708)
Front Neurosci. 2022 Mar 24;16:842635. (PMID: 35401092)
Front Neurorobot. 2017 Aug 15;11:38. (PMID: 28860986)
Front Neuroinform. 2024 Feb 29;18:1345425. (PMID: 38486923)
Sci Rep. 2024 Aug 30;14(1):20237. (PMID: 39215126)
Neurorehabil Neural Repair. 2015 Jun;29(5):462-71. (PMID: 25385765)
IEEE Trans Neural Syst Rehabil Eng. 2020 Dec;28(12):2773-2782. (PMID: 33378260)
IEEE Trans Neural Syst Rehabil Eng. 2024;32:3338-3347. (PMID: 39196743)
IEEE J Biomed Health Inform. 2024 Dec;28(12):7126-7137. (PMID: 39190517)
IEEE Trans Biomed Eng. 2023 Feb;70(2):436-445. (PMID: 35867371)
Alzheimers Res Ther. 2022 Nov 12;14(1):170. (PMID: 36371269)
IEEE Trans Neural Syst Rehabil Eng. 2023;31:3188-3200. (PMID: 37498754)
Front Neurosci. 2015 Nov 18;9:438. (PMID: 26635514)
IEEE Trans Neural Syst Rehabil Eng. 2023;31:1900-1911. (PMID: 37027253)
PLoS One. 2022 Jul 22;17(7):e0268880. (PMID: 35867703)
J Neural Eng. 2024 May 13;21(3):. (PMID: 38684154)
Sci Rep. 2026 Jan 30;16(1):6674. (PMID: 41617819)
Front Neurosci. 2020 Nov 19;14:559858. (PMID: 33328845)
Sci Rep. 2025 Oct 3;15(1):34555. (PMID: 41044308)
Neural Netw. 2020 Sep;129:55-74. (PMID: 32502798)
IEEE Trans Neural Syst Rehabil Eng. 2023;31:710-719. (PMID: 37015413)
J Neural Eng. 2020 Oct 13;17(5):056008. (PMID: 32916675)
J Neural Eng. 2021 Mar 25;18(4):. (PMID: 33691299)
Sensors (Basel). 2018 Aug 29;18(9):. (PMID: 30158505)
Front Neurosci. 2018 Jun 28;12:438. (PMID: 30002615)
Nat Rev Neurosci. 2019 Oct;20(10):624-634. (PMID: 31384033)
Front Hum Neurosci. 2017 May 29;11:274. (PMID: 28611611)
J Neural Eng. 2024 May 20;21(3):. (PMID: 38718788)
Front Neurosci. 2019 Aug 07;13:822. (PMID: 31440129)
Grant Information: No. RS-2025-25419004 National Research Foundation of Korea; No. RS-2024-00397674 National Research Foundation of Korea; IITP-2026-RS-2023-00258971 Ministry of Science and ICT
Contributed Indexing: Keywords: Ear-EEG; brain–computer interface; deep learning architecture; lightweight convolutional neural network; mental imagery
Entry Date(s): Date Created: 20260826 Date Completed: 20260826 Latest Revision: 20260828
Update Code: 20260828
PubMed Central ID: PMC13510306
DOI: 10.3390/bios16080437
PMID: 42645055
Βάση Δεδομένων: MEDLINE
FullText Text:
  Availability: 0
CustomLinks:
  – Url: https://resolver.ebsco.com/c/fiv2js/result?sid=EBSCO:cmedm&genre=article&issn=20796374&ISBN=&volume=16&issue=8&date=20260812&spage=&pages=&title=Biosensors&atitle=Deep%20Learning%20for%20Ear-EEG-Based%20Brain-Computer%20Interface%3A%20A%20Systematic%20Comparison%20and%20Design%20Insights.&aulast=Kim%20JS&id=DOI:10.3390/bios16080437
    Name: Full Text Finder (for New FTF UI) (ns324271)
    Category: fullText
    Text: Full Text Finder
    MouseOverText: Full Text Finder
Header DbId: cmedm
DbLabel: MEDLINE
An: 42645055
AccessLevel: 3
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Deep Learning for Ear-EEG-Based Brain-Computer Interface: A Systematic Comparison and Design Insights.
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AU" term="%22Kim+JS%22">Kim JS</searchLink>; Department of Computer and Information Science, Korea University, Sejong 30019, Republic of Korea.<br /><searchLink fieldCode="AU" term="%22Choi+SI%22">Choi SI</searchLink>; Medical Metrology Group, Division of Biomedical Metrology, Korea Research Institute of Standards and Science (KRISS), Daejeon 34113, Republic of Korea.<br /><searchLink fieldCode="AU" term="%22Hwang+HJ%22">Hwang HJ</searchLink>; Department of Electronics and Information Engineering, Korea University, Sejong 30019, Republic of Korea.; Digital Healthcare Center, Sejong Institute for Business and Technology, Korea University, Sejong 30019, Republic of Korea.; Interdisciplinary Graduate Program for Artificial Intelligence Smart Convergence Technology, Korea University, Sejong 30019, Republic of Korea.<br /><searchLink fieldCode="AU" term="%22Han+CH%22">Han CH</searchLink>; Digital Healthcare Center, Sejong Institute for Business and Technology, Korea University, Sejong 30019, Republic of Korea.; Department of Computer Science and Software Engineering, Korea University, Sejong 30019, Republic of Korea.
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="JN" term="%22101609191%22">Biosensors</searchLink> [Biosensors (Basel)] 2026 Aug 12; Vol. 16 (8). <i>Date of Electronic Publication: </i>2026 Aug 12.
– Name: TypePub
  Label: Publication Type
  Group: TypPub
  Data: Journal Article
– Name: Language
  Label: Language
  Group: Lang
  Data: English
– Name: TitleSource
  Label: Journal Info
  Group: Src
  Data: <i>Publisher: </i><searchLink fieldCode="PB" term="%22MDPI+Pub%22">MDPI Pub </searchLink><i>Country of Publication: </i>Switzerland <i>NLM ID: </i>101609191 <i>Publication Model: </i>Electronic <i>Cited Medium: </i>Internet <i>ISSN: </i>2079-6374 (Electronic) <i>Linking ISSN: </i><searchLink fieldCode="IS" term="%2220796374%22">20796374 </searchLink><i>NLM ISO Abbreviation: </i>Biosensors (Basel) <i>Subsets: </i>MEDLINE
– Name: PublisherInfo
  Label: Imprint Name(s)
  Group: PubInfo
  Data: <i>Original Publication</i>: Basel, Switzerland : MDPI Pub.
– Name: SubjectMESH
  Label: MeSH Terms
  Group: Su
  Data: <searchLink fieldCode="MM" term="%22Brain-Computer+Interfaces%22">Brain-Computer Interfaces*</searchLink> <br /><searchLink fieldCode="MM" term="%22Deep+Learning%22">Deep Learning*</searchLink> <br /><searchLink fieldCode="MM" term="%22Electroencephalography%22">Electroencephalography*</searchLink> <br /><searchLink fieldCode="MM" term="%22Ear%22">Ear*</searchLink><br /><searchLink fieldCode="MH" term="%22Humans%22">Humans</searchLink> ; <searchLink fieldCode="MH" term="%22Neural+Networks%2C+Computer%22">Neural Networks, Computer</searchLink> ; <searchLink fieldCode="MH" term="%22Signal-To-Noise+Ratio%22">Signal-To-Noise Ratio</searchLink>
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Electroencephalography (EEG) measured inside or around ears, called ear-EEG, provides a practical measurement modality for daily brain-computer interface (BCI) applications. However, reliable decoding of mental imagery remains challenging due to the limited number of channels, low signal-to-noise ratio (SNR), and substantial inter- and intra-subject variability inherent to ear-EEG. Addressing these constraints requires advanced decoding strategies specifically optimized for this signal domain. In this study, we retrospectively analyze the ear-EEG dataset of a previous study in which a real-time endogenous BCI was evaluated using conventional machine learning. Specifically, we present an offline benchmark of 23 deep neural network architectures originally developed for scalp-EEG, which were adapted to ear-EEG and evaluated under an identical validation framework. To the best of our knowledge, this is the first systematic comparison of this breadth for ear-EEG-based mental-task classification. Beyond conventional performance comparison, we identify the optimal architecture by jointly considering statistical significance and a performance-cost trade-off, incorporating classification accuracy, parameter count, and measured computational cost. Our results demonstrate that FBLightConvNet achieves the highest classification accuracy among all evaluated models and outperforms common spatial pattern-linear discriminant analysis (CSP-LDA), a widely adopted and robust conventional baseline, on all three recording days, with the difference reaching statistical significance on Days 2 and 3. Notably, many state-of-the-art scalp-EEG models fail to generalize effectively to ear-EEG, highlighting the importance of architecture selection in this domain. These findings identify the best-performing architecture in this setting and indicate which architectural characteristics support effective ear-EEG decoding. Ultimately, this study offers practical design insights and a reproducible benchmarking framework for developing lightweight and high-performance deep learning models, which we hope will support future efforts toward real-world ear-EEG-based BCI systems.
– Name: Ref
  Label: References
  Group: RefInfo
  Data: IEEE Rev Biomed Eng. 2009;2:187-199. (PMID: <searchLink fieldCode="PM" term="%2220442804%22">20442804)</searchLink><br />Front Comput Neurosci. 2023 Apr 26;17:1142948. (PMID: <searchLink fieldCode="PM" term="%2237180880%22">37180880)</searchLink><br />J Neural Eng. 2018 Oct;15(5):056013. (PMID: <searchLink fieldCode="PM" term="%2229932424%22">29932424)</searchLink><br />Hum Brain Mapp. 2017 Nov;38(11):5391-5420. (PMID: <searchLink fieldCode="PM" term="%2228782865%22">28782865)</searchLink><br />Neurology. 2023 Apr 25;100(17):e1750-e1762. (PMID: <searchLink fieldCode="PM" term="%2236878708%22">36878708)</searchLink><br />Front Neurosci. 2022 Mar 24;16:842635. (PMID: <searchLink fieldCode="PM" term="%2235401092%22">35401092)</searchLink><br />Front Neurorobot. 2017 Aug 15;11:38. (PMID: <searchLink fieldCode="PM" term="%2228860986%22">28860986)</searchLink><br />Front Neuroinform. 2024 Feb 29;18:1345425. (PMID: <searchLink fieldCode="PM" term="%2238486923%22">38486923)</searchLink><br />Sci Rep. 2024 Aug 30;14(1):20237. (PMID: <searchLink fieldCode="PM" term="%2239215126%22">39215126)</searchLink><br />Neurorehabil Neural Repair. 2015 Jun;29(5):462-71. (PMID: <searchLink fieldCode="PM" term="%2225385765%22">25385765)</searchLink><br />IEEE Trans Neural Syst Rehabil Eng. 2020 Dec;28(12):2773-2782. (PMID: <searchLink fieldCode="PM" term="%2233378260%22">33378260)</searchLink><br />IEEE Trans Neural Syst Rehabil Eng. 2024;32:3338-3347. (PMID: <searchLink fieldCode="PM" term="%2239196743%22">39196743)</searchLink><br />IEEE J Biomed Health Inform. 2024 Dec;28(12):7126-7137. (PMID: <searchLink fieldCode="PM" term="%2239190517%22">39190517)</searchLink><br />IEEE Trans Biomed Eng. 2023 Feb;70(2):436-445. (PMID: <searchLink fieldCode="PM" term="%2235867371%22">35867371)</searchLink><br />Alzheimers Res Ther. 2022 Nov 12;14(1):170. (PMID: <searchLink fieldCode="PM" term="%2236371269%22">36371269)</searchLink><br />IEEE Trans Neural Syst Rehabil Eng. 2023;31:3188-3200. (PMID: <searchLink fieldCode="PM" term="%2237498754%22">37498754)</searchLink><br />Front Neurosci. 2015 Nov 18;9:438. (PMID: <searchLink fieldCode="PM" term="%2226635514%22">26635514)</searchLink><br />IEEE Trans Neural Syst Rehabil Eng. 2023;31:1900-1911. (PMID: <searchLink fieldCode="PM" term="%2237027253%22">37027253)</searchLink><br />PLoS One. 2022 Jul 22;17(7):e0268880. (PMID: <searchLink fieldCode="PM" term="%2235867703%22">35867703)</searchLink><br />J Neural Eng. 2024 May 13;21(3):. (PMID: <searchLink fieldCode="PM" term="%2238684154%22">38684154)</searchLink><br />Sci Rep. 2026 Jan 30;16(1):6674. (PMID: <searchLink fieldCode="PM" term="%2241617819%22">41617819)</searchLink><br />Front Neurosci. 2020 Nov 19;14:559858. (PMID: <searchLink fieldCode="PM" term="%2233328845%22">33328845)</searchLink><br />Sci Rep. 2025 Oct 3;15(1):34555. (PMID: <searchLink fieldCode="PM" term="%2241044308%22">41044308)</searchLink><br />Neural Netw. 2020 Sep;129:55-74. (PMID: <searchLink fieldCode="PM" term="%2232502798%22">32502798)</searchLink><br />IEEE Trans Neural Syst Rehabil Eng. 2023;31:710-719. (PMID: <searchLink fieldCode="PM" term="%2237015413%22">37015413)</searchLink><br />J Neural Eng. 2020 Oct 13;17(5):056008. (PMID: <searchLink fieldCode="PM" term="%2232916675%22">32916675)</searchLink><br />J Neural Eng. 2021 Mar 25;18(4):. (PMID: <searchLink fieldCode="PM" term="%2233691299%22">33691299)</searchLink><br />Sensors (Basel). 2018 Aug 29;18(9):. (PMID: <searchLink fieldCode="PM" term="%2230158505%22">30158505)</searchLink><br />Front Neurosci. 2018 Jun 28;12:438. (PMID: <searchLink fieldCode="PM" term="%2230002615%22">30002615)</searchLink><br />Nat Rev Neurosci. 2019 Oct;20(10):624-634. (PMID: <searchLink fieldCode="PM" term="%2231384033%22">31384033)</searchLink><br />Front Hum Neurosci. 2017 May 29;11:274. (PMID: <searchLink fieldCode="PM" term="%2228611611%22">28611611)</searchLink><br />J Neural Eng. 2024 May 20;21(3):. (PMID: <searchLink fieldCode="PM" term="%2238718788%22">38718788)</searchLink><br />Front Neurosci. 2019 Aug 07;13:822. (PMID: <searchLink fieldCode="PM" term="%2231440129%22">31440129)</searchLink>
– Name: GrantInfo
  Label: Grant Information
  Group: Grant
  Data: No. RS-2025-25419004 National Research Foundation of Korea; No. RS-2024-00397674 National Research Foundation of Korea; IITP-2026-RS-2023-00258971 Ministry of Science and ICT
– Name: SubjectMinor
  Label: Contributed Indexing
  Group:
  Data: <i>Keywords: </i>Ear-EEG; brain–computer interface; deep learning architecture; lightweight convolutional neural network; mental imagery
– Name: DateEntry
  Label: Entry Date(s)
  Group: Date
  Data: <i>Date Created: </i>20260826 <i>Date Completed: </i>20260826 <i>Latest Revision: </i>20260828
– Name: DateUpdate
  Label: Update Code
  Group: Date
  Data: 20260828
– Name: PubmedCentralID
  Label: PubMed Central ID
  Group: ID
  Data: PMC13510306
– Name: DOI
  Label: DOI
  Group: ID
  Data: 10.3390/bios16080437
– Name: AN
  Label: PMID
  Group: ID
  Data: 42645055
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=cmedm&AN=42645055
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.3390/bios16080437
    Languages:
      – Code: eng
        Text: English
    Subjects:
      – SubjectFull: Humans
        Type: general
      – SubjectFull: Neural Networks, Computer
        Type: general
      – SubjectFull: Signal-To-Noise Ratio
        Type: general
      – SubjectFull: Brain-Computer Interfaces
        Type: general
      – SubjectFull: Deep Learning
        Type: general
      – SubjectFull: Electroencephalography
        Type: general
      – SubjectFull: Ear
        Type: general
    Titles:
      – TitleFull: Deep Learning for Ear-EEG-Based Brain-Computer Interface: A Systematic Comparison and Design Insights.
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Kim JS
      – PersonEntity:
          Name:
            NameFull: Choi SI
      – PersonEntity:
          Name:
            NameFull: Hwang HJ
      – PersonEntity:
          Name:
            NameFull: Han CH
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 12
              M: 08
              Text: 2026 Aug 12
              Type: published
              Y: 2026
          Identifiers:
            – Type: issn-electronic
              Value: 2079-6374
          Numbering:
            – Type: volume
              Value: 16
            – Type: issue
              Value: 8
          Titles:
            – TitleFull: Biosensors
              Type: main
ResultId 1