Development of a Novel Post-Processing Algorithm for Myoelectric Pattern Classification

Masahiro Kasuya, Ryu Kato, Hiroshi Yokoi

    Abstract

    This paper describes a novel post-processing algorithm for electromyographic (EMG) pattern classification, for use with myoelectric prosthetic hands. Amputees have difficulties controllingmultiple degrees of freedom, but there is an increasingnumber of prosthetic hands with multiple degrees of freedom. Generally, an increasingnumber of classes decreases the classification accuracy. Artificial neural networks have been used for EMG pattern classification in previous studies. The proposed post-processingalg orithm stores the temporal sequence of classifications from the EMG pattern classification algorithm, and runs a second classification based on the sequential patterns. We compared the accuracy of the output before and after the post-processingstep. In our experiment, we set the trainingtime of the EMG pattern classification algorithm to 1 s for each class, and used three channels of surface EMG signals. We selected 7 and 9 classes of EMG patterns, and recorded the output every 10-20 ms. The classification accuracy improved by 11.5% with 7 classes, and 17.7% with 9 classes. The overall accuracy of the proposed system was 82.5% for 9 classes and 92.9% for 7 classes. With the adequately high classification accuracy and other features(small number of EMG channels and short trainingtime ), the proposed method is potentially suitable for practical use with prosthetic hands.

    Original languageEnglish
    Pages (from-to)217-224
    Number of pages8
    JournalTransactions of Japanese Society for Medical and Biological Engineering
    Volume53
    Issue number4
    DOIs
    StatePublished - 2015 Dec 10

    Fingerprint

    Pattern recognition
    Prosthetics
    Neural networks
    Experiments

    Keywords

    • Biomedical engineering
    • Classification
    • Cybernetics
    • Myoelectric
    • Prosthetic hand

    ASJC Scopus subject areas

    • Biomedical Engineering

    Cite this

    Development of a Novel Post-Processing Algorithm for Myoelectric Pattern Classification. / Kasuya, Masahiro; Kato, Ryu; Yokoi, Hiroshi.

    In: Transactions of Japanese Society for Medical and Biological Engineering, Vol. 53, No. 4, 10.12.2015, p. 217-224.

    Research output: Contribution to journalArticle

    Kasuya, Masahiro; Kato, Ryu; Yokoi, Hiroshi / Development of a Novel Post-Processing Algorithm for Myoelectric Pattern Classification.

    In: Transactions of Japanese Society for Medical and Biological Engineering, Vol. 53, No. 4, 10.12.2015, p. 217-224.

    Research output: Contribution to journalArticle

    @article{6c4b5d9436e2405db1dd8002cd7864d1,
    title = "Development of a Novel Post-Processing Algorithm for Myoelectric Pattern Classification",
    keywords = "Biomedical engineering, Classification, Cybernetics, Myoelectric, Prosthetic hand",
    author = "Masahiro Kasuya and Ryu Kato and Hiroshi Yokoi",
    year = "2015",
    month = "12",
    doi = "10.11239/ajsmbe.53.217",
    volume = "53",
    pages = "217--224",
    journal = "Japanese Journal of Medical Electronics and Biological Engineering",
    issn = "1347-443X",
    publisher = "Nihon M E Gakkai",
    number = "4",

    }

    TY - JOUR

    T1 - Development of a Novel Post-Processing Algorithm for Myoelectric Pattern Classification

    AU - Kasuya,Masahiro

    AU - Kato,Ryu

    AU - Yokoi,Hiroshi

    PY - 2015/12/10

    Y1 - 2015/12/10

    N2 - This paper describes a novel post-processing algorithm for electromyographic (EMG) pattern classification, for use with myoelectric prosthetic hands. Amputees have difficulties controllingmultiple degrees of freedom, but there is an increasingnumber of prosthetic hands with multiple degrees of freedom. Generally, an increasingnumber of classes decreases the classification accuracy. Artificial neural networks have been used for EMG pattern classification in previous studies. The proposed post-processingalg orithm stores the temporal sequence of classifications from the EMG pattern classification algorithm, and runs a second classification based on the sequential patterns. We compared the accuracy of the output before and after the post-processingstep. In our experiment, we set the trainingtime of the EMG pattern classification algorithm to 1 s for each class, and used three channels of surface EMG signals. We selected 7 and 9 classes of EMG patterns, and recorded the output every 10-20 ms. The classification accuracy improved by 11.5% with 7 classes, and 17.7% with 9 classes. The overall accuracy of the proposed system was 82.5% for 9 classes and 92.9% for 7 classes. With the adequately high classification accuracy and other features(small number of EMG channels and short trainingtime ), the proposed method is potentially suitable for practical use with prosthetic hands.

    AB - This paper describes a novel post-processing algorithm for electromyographic (EMG) pattern classification, for use with myoelectric prosthetic hands. Amputees have difficulties controllingmultiple degrees of freedom, but there is an increasingnumber of prosthetic hands with multiple degrees of freedom. Generally, an increasingnumber of classes decreases the classification accuracy. Artificial neural networks have been used for EMG pattern classification in previous studies. The proposed post-processingalg orithm stores the temporal sequence of classifications from the EMG pattern classification algorithm, and runs a second classification based on the sequential patterns. We compared the accuracy of the output before and after the post-processingstep. In our experiment, we set the trainingtime of the EMG pattern classification algorithm to 1 s for each class, and used three channels of surface EMG signals. We selected 7 and 9 classes of EMG patterns, and recorded the output every 10-20 ms. The classification accuracy improved by 11.5% with 7 classes, and 17.7% with 9 classes. The overall accuracy of the proposed system was 82.5% for 9 classes and 92.9% for 7 classes. With the adequately high classification accuracy and other features(small number of EMG channels and short trainingtime ), the proposed method is potentially suitable for practical use with prosthetic hands.

    KW - Biomedical engineering

    KW - Classification

    KW - Cybernetics

    KW - Myoelectric

    KW - Prosthetic hand

    UR - http://www.scopus.com/inward/record.url?scp=84952063345&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84952063345&partnerID=8YFLogxK

    U2 - 10.11239/ajsmbe.53.217

    DO - 10.11239/ajsmbe.53.217

    M3 - Article

    VL - 53

    SP - 217

    EP - 224

    JO - Japanese Journal of Medical Electronics and Biological Engineering

    T2 - Japanese Journal of Medical Electronics and Biological Engineering

    JF - Japanese Journal of Medical Electronics and Biological Engineering

    SN - 1347-443X

    IS - 4

    ER -