Authors: R. Janaki; S. Nirmala Devi
Addresses: Anna University, Chennai, India ' Anna University, Chennai, India
Abstract: The dermoscopic images of skin lesions implicitly express the lesion characteristics. The study of salient features of the dermoscopic images would be highly helpful in the categorisation of skin lesions. The critical task is to assort the skin lesion as cancerous (melanoma) and non-cancerous (non-melanoma). Before salient features extraction, the lesion has to be carved up from the healthy skin region. It is required that the lesion has to be segmented accurately and automatically to facilitate delineation of the skin lesions in the form of key features. In this paper, an attempt has been made to form the feature vector of the segmented image which can be used for effectively classifying the dermoscopic image as melanoma or not.
Keywords: study of dermoscopic images; dermoscopic image analysis; feature extraction of dermoscopic images; automatic edge-based segmentation of dermoscopic images; melanoma detection; the anisotropic diffusion filter; level set function; image segmentation evaluation; local binary patterns; HSV colour histogram features; LTP; LTrP.
International Journal of Biomedical Engineering and Technology, 2017 Vol.24 No.4, pp.393 - 409
Received: 30 May 2016
Accepted: 07 Jul 2016
Published online: 05 Jul 2017 *