• DocumentCode
    600186
  • Title

    Capability of reflectance features in diagnosing colorectal polyps using white light colonoscopy

  • Author

    Wei-Chih Shen ; Chun-Lung Feng

  • Author_Institution
    Comput. Sci. & Inf. Eng., Asia Univ., Taichung, Taiwan
  • fYear
    2012
  • fDate
    4-7 Nov. 2012
  • Firstpage
    574
  • Lastpage
    578
  • Abstract
    In white light colonoscopy (WLC), three tissue illuminations split from full spectrum of visible spectrum characterize the colorectal mucosa as three reflectance data sets from the superficial surface to the depth of light penetrating. Using reflectance features defined on each data set rather than on superimposed data sets, the diagnostic capabilities of WLC without magnification in differentiating adenoma from hyperplasia polyps were evaluated and compared. We retrospectively collected 89 polyps from 82 patients who underwent the colonoscopy using WLC in November 2011. Of all enrolled polyps, the pathology results included 46 hyperplastic polyps and 43 adenomas. A computer-aided diagnosis (CAD) system was established to identify, characterize, and predict malignancy using the most descriptive factors identified from all reflectance features. The diagnostic performance of CAD system using WLC was accuracy 85.4% (76/89), sensitivity 86% (37/43), and specificity 84.8% (39/46). Using reflectance features, the diagnostic capabilities of WLC demonstrated significant diagnostic performances to objectively provide auxiliary information for physicians in predicting adenoma.
  • Keywords
    biomedical optical imaging; cancer; computer aided analysis; endoscopes; feature extraction; medical image processing; reflectivity; sensitivity; tumours; adenoma polyps; auxiliary information; colorectal mucosa; colorectal polyps diagnosis; computer-aided diagnosis system; diagnostic capability; hyperplasia polyps; light penetrating depth; malignancy; pathology; physicians; reflectance data sets; reflectance feature capability; sensitivity; superficial surface; superimposed data sets; tissue illuminations; visible spectrum; white light colonoscopy; Colonoscopy; Design automation; Entropy; Green products; Imaging; Lighting; Reflectivity; Colorectal Cancer; Computer-Aided Diagnosis; Reflectance Feature; White Light Colonoscopy;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Intelligent Signal Processing and Communications Systems (ISPACS), 2012 International Symposium on
  • Conference_Location
    New Taipei
  • Print_ISBN
    978-1-4673-5083-9
  • Electronic_ISBN
    978-1-4673-5081-5
  • Type

    conf

  • DOI
    10.1109/ISPACS.2012.6473555
  • Filename
    6473555