05984810

a

05984810 Ciecholewski, Marcin AdaBoost-based approach for detecting lithiasis and polyps in USG images of the gallbladder. Badioze Zaman, Halimah (ed.) et al., Visual informatics: sustaining research and innovations. Second international visual informatics conference, IVIC 2011, Selangor, Malaysia, November 9--11, 2011. Proceedings, Part I. Berlin: Springer (ISBN 978-3-642-25190-0/pbk). Lecture Notes in Computer Science 7066, 206-215 (2011). 2011 Berlin: Springer EN adaptive boosting classification medical image analysis gallbladder lesions ultrasonography (USG) visual informatics

doi:10.1007/978-3-642-25191-7_20

Summary: This article presents the application of the AdaBoost method to recognise gallbladder lesions such as lithiasis and polyps in USG images. The classifier handles rectangular input image areas of a specific length. If the diameter of areas segmented is much greater than the diameter expected on the input, wavelet approximation of input images is used. The classification results obtained by using the AdaBoost method are promising for lithiasis classification. In the best case, the algorithm achieved the accuracy of 91\% for lithiasis and of 80\% when classifyingpolyps, as well as the accuracy of 78.9\% for polyps and lithiasis jointly.