Organized Classification of Melanoma Images Using Gaussian Mixture Model and Artificial Neural Network
, and .
Imperial Journal of Interdisciplinary Research (2017)

The most dangerous form of skin cancer ,tese cancerous growths develop when unrepired DNA damage to the skin cells genetic defects that lead the ski cells to multiply rapidy and form maligant tumors .In melanoma diagnosis, the intersect and ascertain detection of the melanocytes in the epidermis area is an important step before the diagnosis is made. If the melanocytes can be found correctly, architectural and cellular features e.g. size, distribution, location etc. they can be used to grade or determine the malignancy of the skin tissue. So border detection of the infected regions in this automated system is the major step as all other results depends on accuracy of detected area. In this work, we have proposed a neural network based segmentation system which can efficiently classify the pixels of the input images into infected or non-infected parts. In this we used the pixel surroundings which contain the neighborhood texture information of the pixels which helps in better classification of pixels into similar clusters depending upon the surroundings. For getting the features of theses neighborhood Gaussian mixtures has been evaluated first from a few selected pixels from different intensity areas and then each pixel has been put into a particular cluster based on modeling of Gaussian mixtures. For further improving the segmentation results and to get results in two categories, neural network based classification has been preferred as they gave much better efficiency than the pre-segmentation done by Gaussian clustering. The proposed work has been tested on variety of images and the results shows that it gives higher accuracy in terms of sensitivity and specificity values.
  • @ijirjournal
This publication has not been reviewed yet.

rating distribution
average user rating0.0 out of 5.0 based on 0 reviews
    Please log in to take part in the discussion (add own reviews or comments).