A robust approach for automatic skin cancer disease classification

Skin cancers particularly malignant melanoma is lethal and difficult to identify in last stages. The variation of stages (Squamous Cell Carcinoma, Actinic Keratosis Cell, Basal cell and Malignant Melanoma) of Skin Cancer is highly ambiguous and difficult to recognize. To clearly recognize the stage of Skin Cancer is primarily important for effective treatment, which cause in increasing the survival rate from Skin cancer. In this work, we propose a methodology to reduce the probability of false diagnosis. In the proposed methodology, the data set is first preprocessed using K-mean Clustering algorithm. This preprocessing helps to increase the rate of reorganization by removing all irrelevant texture. The preprocessed data is then used to extract the features. The classification results illustrate that the proposed method can considerably improve in classification of Skin cancer disease. The computed accuracy of classification for this algorithm is achieved up to 94.4%.