الفهرس 
Title : A graph theory approach towards melanoma detection
SummaryOnly 14 pages are availabe for public view
 |  | from | 99 |  |  |
| | | from | 99 | | |
Abstract
Melanoma is the most fatal type of skin cancer. Detection of melanoma from dermoscopic images in an early stage is critical for improving survival rates. Previous studies show that the detection performance depends significantly on the skin lesion image representations and features. In this work, we propose a melanoma detection approach that combines graph-theoretic representations with conventional dermoscopic image features to enhance the detection performance. A superpixel graph is constructed by generating supepixels for the dermoscopic images. An edge of such a graph connects two adjacent superpixels. Features are extracted from different graph models in the vertex domain at both local and global scales and in the spectral domain using the graph Fourier transform (GFT). Other features based on color, geometry, and texture are extracted from the original images. Datasets from the international skin imaging collaboration (ISIC) archive is fed to the system which achieved an AUC of 99.91%, an accuracy of 97.4%, a specificity of 95.16% and a sensitivity of 100%