GLOBAL PROCESSING SYSTEM BASED SKIN CANCER CLASSIFICATION SING DERMOSCOPIC IMAGES
European Journal of Molecular & Clinical Medicine,
2020, Volume 7, Issue 10, Pages 2441-2447
AbstractThe Global Processing System (GPS) of Non-Subsampled Shearlet Transform (NSST) features for dermoscopic image classification with Support Vector Machine (SVM) is presented. If a skin cancer is diagnosed early in its development, i.e., when the tumour is thin, it has a good prognosis which significantly worsens as the thickness increases. The NSST is decomposed by 4 levels with 8 directions. Finally, the SVM classifier is used for classification. The proposed system produces the classification accuracy of 96 % and its sensitivity is 93.33 % and specificity 100 %.
- Article View: 72
- PDF Download: 166