2021/07/26 by Debesh Jha, Jha, Debesh, Pia H. Smedsrud +11 · 4 citations
Computer Science · Medicine · #Colorectal Cancer Screening and Detection #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Image Retrieval and Classification Techniques #Radiomics and Machine Learning in Medical Imaging
paper · pdf · doi:10.48550/arxiv.2107.12435
openalex publication_date 2021/07/26 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
Colonoscopy is considered the gold standard for detection of colorectal\ncancer and its precursors. Existing examination methods are, however, hampered\nby high overall miss-rate, and many abnormalities are left undetected.\nComputer-Aided Diagnosis systems based on advanced machine learning algorithms\nare touted as a game-changer that can identify regions in the colon overlooked\nby the physicians during endoscopic examinations, and help detect and\ncharacterize lesions. In previous work, we have proposed the ResUNet++\narchitecture and demonstrated that it produces more efficient results compared\nwith its counterparts U-Net and ResUNet. In this paper, we demonstrate that\nfurther improvements to the overall prediction performance of the ResUNet++\narchitecture can be achieved by using conditional random field and test-time\naugmentation. We have performed extensive evaluations and validated the\nimprovements using six publicly available datasets: Kvasir-SEG, CVC-ClinicDB,\nCVC-ColonDB, ETIS-Larib Polyp DB, ASU-Mayo Clinic Colonoscopy Video Database,\nand CVC-VideoClinicDB. Moreover, we compare our proposed architecture and\nresulting model with other State-of-the-art methods. To explore the\ngeneralization capability of ResUNet++ on different publicly available polyp\ndatasets, so that it could be used in a real-world setting, we performed an\nextensive cross-dataset evaluation. The experimental results show that applying\nCRF and TTA improves the performance on various polyp segmentation datasets\nboth on the same dataset and cross-dataset.\n