Clustering-Based Melanoma Detection in Dermoscopy Images Using ABCD Parameters

Melanoma, dangerous among skin cancer, becomes fatal when not diagnosed and treated at the earliest. It can be correctly predicted only by the expert dermatologists. Owing to lack of experts, computer-aided diagnosis is preferred nowadays. Here we have proposed the image processing algorithm based on clustering to identify melanoma. A total of 170 images taken from the standard database are used to test the algorithm. Various filters and pre-processing techniques have been analyzed for better skin enhancement. The lesion portion is segmented using K-means clustering algorithm. Then the features are extracted from the segmented lesion, and total dermatoscopy score was calculated. This score was calculated for all images and are classified into melanoma and non-melanoma. Finally, the classification accuracy of the algorithm is computed.