Classification of various colon diseases in Colonoscopy video using Cross-Wavelet features

One of the leading cause of deaths from cancer being colon cancer, which is observed in 10% of men and 9.2% in case of women [1], prevention technique has been approached where the major chance lies in detection in early stage during Optical Colonoscopy (OC) of the subject. This paper proposes a novel scheme for detecting various colon diseases through Colonoscopy videos using Cross-Wavelet Transform (XWT) [2]. The datasets consist of five different colonoscopy videos which are split into five sets of image datasets i.e., Colon Polyp, Colon Ulcer, Chron's Diseases, Sigmoidal Colon and Normal Colon collected from American College of Gastroenterology [3], Medico Gastroenterologo [4] and Gastrolab [5]. A novel cross-wavelet transform, is opted for extracting the features from the datasets. This is followed by a multiclass classification using Multiclass Support Vector Machine (MSVM). The recognition rate using this improved technique (XWT_MSVM) is 98.46% which is far better than any other well-known and benchmark procedures.