SEGCROP: Segmentation-based dynamic cropping of endoscopic videos to address label leakage in surgical tool detection

Adnan Qayyum, Muhammad Bilal, Junaid Qadir, Massimo Caputo, Hunaid Vohra, Taofeek Akinosho, Ilhem Berrou, Faatihah Niyi-Odumosu, Michael Loizou, Anuoluwapo Ajayi, Sofiat Abioye

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

    58 Downloads (Pure)

    Abstract

    In recent times, surgical data science has emerged as an important research discipline in interventional healthcare. There
    are many potential applications for analysing endoscopic surgical videos using machine learning (ML) techniques such
    as surgical tool classification, action recognition, and tissue segmentation. However, the efficacy of ML algorithms to learn robust features drastically deteriorates when models are trained on noise-affected data [1]. Appropriate data preprocessing for endoscopic videos is thus crucial to ensure robust ML training. To this end, we demonstrate the presence of label leakage when surgical tool classification is performed naively and present SegCrop, a dynamic U-Net model with an integrated attention mechanism to dynamically crop the arbitrary field of view (FoV) in endoscopic surgical videos to remove spurious label-related information from the data. In addition, we leverage explainability techniques to demonstrate how the presence of spurious correlations influences the
    model’s learning capability.
    Original languageEnglish
    Title of host publicationIEEE International Symposium on Biomedical Imaging (ISBI)
    Publication statusUnpublished - 2023
    EventIEEE International Symposium on Biomedical Imaging - Cartagena de Indias, Colombia
    Duration: 18 Apr 202323 Apr 2023
    http://2023.biomedicalimaging.org/en/VENUE.html

    Conference

    ConferenceIEEE International Symposium on Biomedical Imaging
    Abbreviated titleISBI 2023
    Country/TerritoryColombia
    CityCartagena de Indias
    Period18/04/2323/04/23
    Internet address

    Fingerprint

    Dive into the research topics of 'SEGCROP: Segmentation-based dynamic cropping of endoscopic videos to address label leakage in surgical tool detection'. Together they form a unique fingerprint.

    Cite this