Segmentation of hand radiographs by using multi-level connected active appearance models

J.A. Kauffman, Cornelis H. Slump, Hein J. Bernelot Moens

    Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

    10 Citations (Scopus)

    Abstract

    Robust and accurate segmentation methods are important for the computerized evaluation of medical images. For treatment of rheumatoid arthritis, joint damage assessment in radiographs of hands is frequently used for monitoring disease progression. Current clinical scoring methods are based on visual measurements that are time-consuming and subject to intra and inter-reader variance. A solution may be found in the development of partially automated assessment procedures. This requires reliable segmentation algorithms.Our work demonstrates a segmentation method based on multiple connected active appearance models (AAM) with multiple search steps using different quality levels. The quality level can be regulated by setting the image resolution and the number of landmarks in the AAMs.We performed experiments using two models of different quality levels for shape and texture information. Both models included AAMs for the carpal region, the metacarpals, and all phalanges. By starting an iterative search with the faster, low-quality model, we were able to determine the initial parameters of the second, high-quality model. After the second search, the results showed successful segmentation for 22 of 30 test images. For these images, 70% of the landmarks were found within 1.3 mm difference from manual placement by an expert.The multi-level search approach resulted in a reduction of 50% in calculation time compared to a search using a single model. Results are expected to improve when the model is refined by increasing the number of training examples and the resolution of the models.
    Original languageEnglish
    Title of host publicationMedical Imaging Conference 2005 MIC
    EditorsJ. Michael Fitzpatrick, Joseph M. Reinhardt
    Place of PublicationBellingham
    PublisherSPIE
    Pages1571-1581
    Number of pages11
    ISBN (Print)0-8194-5721-3
    DOIs
    Publication statusPublished - 5 May 2005
    EventMedical Imaging Conference, MIC 2005 - Town and Country Hotel, San Diego, United States
    Duration: 12 Feb 200517 Feb 2005

    Publication series

    NameProceedings of SPIE
    PublisherSPIE
    Volume5747
    ISSN (Print)0277-786X

    Conference

    ConferenceMedical Imaging Conference, MIC 2005
    Abbreviated titleMIC
    CountryUnited States
    CitySan Diego
    Period12/02/0517/02/05
    Other12-17 Feb 2005

    Fingerprint

    Image resolution
    Textures
    Monitoring
    Experiments

    Keywords

    • CR-I.4

    Cite this

    Kauffman, J. A., Slump, C. H., & Bernelot Moens, H. J. (2005). Segmentation of hand radiographs by using multi-level connected active appearance models. In J. M. Fitzpatrick, & J. M. Reinhardt (Eds.), Medical Imaging Conference 2005 MIC (pp. 1571-1581). (Proceedings of SPIE; Vol. 5747). Bellingham: SPIE. https://doi.org/10.1117/12.595640
    Kauffman, J.A. ; Slump, Cornelis H. ; Bernelot Moens, Hein J. / Segmentation of hand radiographs by using multi-level connected active appearance models. Medical Imaging Conference 2005 MIC. editor / J. Michael Fitzpatrick ; Joseph M. Reinhardt. Bellingham : SPIE, 2005. pp. 1571-1581 (Proceedings of SPIE).
    @inproceedings{61388704f4a54355b671155633a904b9,
    title = "Segmentation of hand radiographs by using multi-level connected active appearance models",
    abstract = "Robust and accurate segmentation methods are important for the computerized evaluation of medical images. For treatment of rheumatoid arthritis, joint damage assessment in radiographs of hands is frequently used for monitoring disease progression. Current clinical scoring methods are based on visual measurements that are time-consuming and subject to intra and inter-reader variance. A solution may be found in the development of partially automated assessment procedures. This requires reliable segmentation algorithms.Our work demonstrates a segmentation method based on multiple connected active appearance models (AAM) with multiple search steps using different quality levels. The quality level can be regulated by setting the image resolution and the number of landmarks in the AAMs.We performed experiments using two models of different quality levels for shape and texture information. Both models included AAMs for the carpal region, the metacarpals, and all phalanges. By starting an iterative search with the faster, low-quality model, we were able to determine the initial parameters of the second, high-quality model. After the second search, the results showed successful segmentation for 22 of 30 test images. For these images, 70{\%} of the landmarks were found within 1.3 mm difference from manual placement by an expert.The multi-level search approach resulted in a reduction of 50{\%} in calculation time compared to a search using a single model. Results are expected to improve when the model is refined by increasing the number of training examples and the resolution of the models.",
    keywords = "CR-I.4",
    author = "J.A. Kauffman and Slump, {Cornelis H.} and {Bernelot Moens}, {Hein J.}",
    year = "2005",
    month = "5",
    day = "5",
    doi = "10.1117/12.595640",
    language = "English",
    isbn = "0-8194-5721-3",
    series = "Proceedings of SPIE",
    publisher = "SPIE",
    pages = "1571--1581",
    editor = "Fitzpatrick, {J. Michael} and Reinhardt, {Joseph M.}",
    booktitle = "Medical Imaging Conference 2005 MIC",
    address = "United States",

    }

    Kauffman, JA, Slump, CH & Bernelot Moens, HJ 2005, Segmentation of hand radiographs by using multi-level connected active appearance models. in JM Fitzpatrick & JM Reinhardt (eds), Medical Imaging Conference 2005 MIC. Proceedings of SPIE, vol. 5747, SPIE, Bellingham, pp. 1571-1581, Medical Imaging Conference, MIC 2005, San Diego, United States, 12/02/05. https://doi.org/10.1117/12.595640

    Segmentation of hand radiographs by using multi-level connected active appearance models. / Kauffman, J.A.; Slump, Cornelis H.; Bernelot Moens, Hein J.

    Medical Imaging Conference 2005 MIC. ed. / J. Michael Fitzpatrick; Joseph M. Reinhardt. Bellingham : SPIE, 2005. p. 1571-1581 (Proceedings of SPIE; Vol. 5747).

    Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

    TY - GEN

    T1 - Segmentation of hand radiographs by using multi-level connected active appearance models

    AU - Kauffman, J.A.

    AU - Slump, Cornelis H.

    AU - Bernelot Moens, Hein J.

    PY - 2005/5/5

    Y1 - 2005/5/5

    N2 - Robust and accurate segmentation methods are important for the computerized evaluation of medical images. For treatment of rheumatoid arthritis, joint damage assessment in radiographs of hands is frequently used for monitoring disease progression. Current clinical scoring methods are based on visual measurements that are time-consuming and subject to intra and inter-reader variance. A solution may be found in the development of partially automated assessment procedures. This requires reliable segmentation algorithms.Our work demonstrates a segmentation method based on multiple connected active appearance models (AAM) with multiple search steps using different quality levels. The quality level can be regulated by setting the image resolution and the number of landmarks in the AAMs.We performed experiments using two models of different quality levels for shape and texture information. Both models included AAMs for the carpal region, the metacarpals, and all phalanges. By starting an iterative search with the faster, low-quality model, we were able to determine the initial parameters of the second, high-quality model. After the second search, the results showed successful segmentation for 22 of 30 test images. For these images, 70% of the landmarks were found within 1.3 mm difference from manual placement by an expert.The multi-level search approach resulted in a reduction of 50% in calculation time compared to a search using a single model. Results are expected to improve when the model is refined by increasing the number of training examples and the resolution of the models.

    AB - Robust and accurate segmentation methods are important for the computerized evaluation of medical images. For treatment of rheumatoid arthritis, joint damage assessment in radiographs of hands is frequently used for monitoring disease progression. Current clinical scoring methods are based on visual measurements that are time-consuming and subject to intra and inter-reader variance. A solution may be found in the development of partially automated assessment procedures. This requires reliable segmentation algorithms.Our work demonstrates a segmentation method based on multiple connected active appearance models (AAM) with multiple search steps using different quality levels. The quality level can be regulated by setting the image resolution and the number of landmarks in the AAMs.We performed experiments using two models of different quality levels for shape and texture information. Both models included AAMs for the carpal region, the metacarpals, and all phalanges. By starting an iterative search with the faster, low-quality model, we were able to determine the initial parameters of the second, high-quality model. After the second search, the results showed successful segmentation for 22 of 30 test images. For these images, 70% of the landmarks were found within 1.3 mm difference from manual placement by an expert.The multi-level search approach resulted in a reduction of 50% in calculation time compared to a search using a single model. Results are expected to improve when the model is refined by increasing the number of training examples and the resolution of the models.

    KW - CR-I.4

    U2 - 10.1117/12.595640

    DO - 10.1117/12.595640

    M3 - Conference contribution

    SN - 0-8194-5721-3

    T3 - Proceedings of SPIE

    SP - 1571

    EP - 1581

    BT - Medical Imaging Conference 2005 MIC

    A2 - Fitzpatrick, J. Michael

    A2 - Reinhardt, Joseph M.

    PB - SPIE

    CY - Bellingham

    ER -

    Kauffman JA, Slump CH, Bernelot Moens HJ. Segmentation of hand radiographs by using multi-level connected active appearance models. In Fitzpatrick JM, Reinhardt JM, editors, Medical Imaging Conference 2005 MIC. Bellingham: SPIE. 2005. p. 1571-1581. (Proceedings of SPIE). https://doi.org/10.1117/12.595640