PET-Train: Automatic Ground Truth Generation from PET Acquisitions for Urinary Bladder Segmentation in CT Images using Deep Learning

Christina Gsaxner, Birgit Pfarrkirchner, Lydia Lindner, Antonio Pepe, Peter M. Roth, Jürgen Wallner, Jan Egger

Publikation: KonferenzbeitragPaperForschungBegutachtung

Abstract

In this contribution, we propose an automatic ground truth generation approach that utilizes Positron Emission Tomography (PET) acquisitions to train neural networks for automatic urinary bladder segmentation in Computed Tomography (CT) images. We evaluated different deep learning architectures to segment the urinary bladder. However, deep neural networks require a large amount of training data, which is currently the main bottleneck in the medical field, because ground truth labels have to be created by medical experts on a time-consuming slice-by-slice basis. To overcome this problem, we generate the training data set from the PET data of combined PET/CT acquisitions. This can be achieved by applying simple thresholding to the PET data, where the radiotracer accumulates very distinct in the urinary bladder. However, the ultimate goal is to entirely skip PET imaging and its additional radiation exposure in the future, and only use CT images for segmentation.

Originalspracheenglisch
DOIs
PublikationsstatusVeröffentlicht - 2019
VeranstaltungIEEE Biomedical Engineering International Conference - Chiang Mai, Thailand
Dauer: 21 Nov 2018 → …
Konferenznummer: 11

Konferenz

KonferenzIEEE Biomedical Engineering International Conference
KurztitelBMEiCON
LandThailand
OrtChiang Mai
Zeitraum21/11/18 → …

Fingerprint

Positron emission tomography
ground truth
bladder
learning
Tomography
positrons
acquisition
tomography
education
Labels
Deep learning
radiation dosage
Neural networks
Imaging techniques
Radiation

Schlagwörter

    ASJC Scopus subject areas

    • Artificial intelligence
    • !!Instrumentation
    • !!Biomedical Engineering

    Dies zitieren

    PET-Train: Automatic Ground Truth Generation from PET Acquisitions for Urinary Bladder Segmentation in CT Images using Deep Learning. / Gsaxner, Christina; Pfarrkirchner, Birgit; Lindner, Lydia; Pepe, Antonio; Roth, Peter M.; Wallner, Jürgen; Egger, Jan.

    2019. Beitrag in IEEE Biomedical Engineering International Conference, Chiang Mai, Thailand.

    Publikation: KonferenzbeitragPaperForschungBegutachtung

    Gsaxner C, Pfarrkirchner B, Lindner L, Pepe A, Roth PM, Wallner J et al. PET-Train: Automatic Ground Truth Generation from PET Acquisitions for Urinary Bladder Segmentation in CT Images using Deep Learning. 2019. Beitrag in IEEE Biomedical Engineering International Conference, Chiang Mai, Thailand. https://doi.org/10.1109/BMEiCON.2018.8609954
    Gsaxner, Christina ; Pfarrkirchner, Birgit ; Lindner, Lydia ; Pepe, Antonio ; Roth, Peter M. ; Wallner, Jürgen ; Egger, Jan. / PET-Train: Automatic Ground Truth Generation from PET Acquisitions for Urinary Bladder Segmentation in CT Images using Deep Learning. Beitrag in IEEE Biomedical Engineering International Conference, Chiang Mai, Thailand.
    @conference{d00b0549a59b4c55985adbdb552a03f9,
    title = "PET-Train: Automatic Ground Truth Generation from PET Acquisitions for Urinary Bladder Segmentation in CT Images using Deep Learning",
    abstract = "In this contribution, we propose an automatic ground truth generation approach that utilizes Positron Emission Tomography (PET) acquisitions to train neural networks for automatic urinary bladder segmentation in Computed Tomography (CT) images. We evaluated different deep learning architectures to segment the urinary bladder. However, deep neural networks require a large amount of training data, which is currently the main bottleneck in the medical field, because ground truth labels have to be created by medical experts on a time-consuming slice-by-slice basis. To overcome this problem, we generate the training data set from the PET data of combined PET/CT acquisitions. This can be achieved by applying simple thresholding to the PET data, where the radiotracer accumulates very distinct in the urinary bladder. However, the ultimate goal is to entirely skip PET imaging and its additional radiation exposure in the future, and only use CT images for segmentation.",
    keywords = "Deep Learning, Medical Imaging, PET/CT, Segmentation, Urinary Bladder",
    author = "Christina Gsaxner and Birgit Pfarrkirchner and Lydia Lindner and Antonio Pepe and Roth, {Peter M.} and J{\"u}rgen Wallner and Jan Egger",
    year = "2019",
    doi = "10.1109/BMEiCON.2018.8609954",
    language = "English",
    note = "IEEE Biomedical Engineering International Conference, BMEiCON ; Conference date: 21-11-2018",

    }

    TY - CONF

    T1 - PET-Train: Automatic Ground Truth Generation from PET Acquisitions for Urinary Bladder Segmentation in CT Images using Deep Learning

    AU - Gsaxner, Christina

    AU - Pfarrkirchner, Birgit

    AU - Lindner, Lydia

    AU - Pepe, Antonio

    AU - Roth, Peter M.

    AU - Wallner, Jürgen

    AU - Egger, Jan

    PY - 2019

    Y1 - 2019

    N2 - In this contribution, we propose an automatic ground truth generation approach that utilizes Positron Emission Tomography (PET) acquisitions to train neural networks for automatic urinary bladder segmentation in Computed Tomography (CT) images. We evaluated different deep learning architectures to segment the urinary bladder. However, deep neural networks require a large amount of training data, which is currently the main bottleneck in the medical field, because ground truth labels have to be created by medical experts on a time-consuming slice-by-slice basis. To overcome this problem, we generate the training data set from the PET data of combined PET/CT acquisitions. This can be achieved by applying simple thresholding to the PET data, where the radiotracer accumulates very distinct in the urinary bladder. However, the ultimate goal is to entirely skip PET imaging and its additional radiation exposure in the future, and only use CT images for segmentation.

    AB - In this contribution, we propose an automatic ground truth generation approach that utilizes Positron Emission Tomography (PET) acquisitions to train neural networks for automatic urinary bladder segmentation in Computed Tomography (CT) images. We evaluated different deep learning architectures to segment the urinary bladder. However, deep neural networks require a large amount of training data, which is currently the main bottleneck in the medical field, because ground truth labels have to be created by medical experts on a time-consuming slice-by-slice basis. To overcome this problem, we generate the training data set from the PET data of combined PET/CT acquisitions. This can be achieved by applying simple thresholding to the PET data, where the radiotracer accumulates very distinct in the urinary bladder. However, the ultimate goal is to entirely skip PET imaging and its additional radiation exposure in the future, and only use CT images for segmentation.

    KW - Deep Learning

    KW - Medical Imaging

    KW - PET/CT

    KW - Segmentation

    KW - Urinary Bladder

    UR - http://www.scopus.com/inward/record.url?scp=85062055886&partnerID=8YFLogxK

    U2 - 10.1109/BMEiCON.2018.8609954

    DO - 10.1109/BMEiCON.2018.8609954

    M3 - Paper

    ER -