Detection of melanoma skin cancer with fine-tuned weight parameters and features using Radial Basis function Network

Abstract

Melanoma Skin cancer is one of the dangerous cancer as well as widespread disease. Worldwide there is 53% melanoma cancer cases, and mortality rate is also increased in the further coming decades.  Detection of skin cancer at the early stages, we can help the human beings form the death rate. To detect cancer at early stage we need to require a computer aided diagnosis mechanism which may help to the dermatologists for accelerating their diagnosis. Here, we propose a computer aided mechanism for detecting melanoma skin cancer with the weights are fine-tuned and extracted the features from the skin cancer image, and we develop the model by using radial basis network. The proposed model initially manages the converting color images to gray scale images by applying the median filter. The main objective of this filter is to reduce the image noise and other unrelated objects from the image. To classify the melanoma skin cancer we have to apply different phases like preprocessing, segmentation, feature extraction and classification. Here, we used PH2 data set for the proposed model and it generate specificity as 86.54%, accuracy as 88.99%, and sensitivity as 91.65%. The results and performance of the model is discussed in the result phase.