Publications

Malignant melanoma has been seen as one of the most precarious form of human cancer. The detection of skin cancer in early stage can be helpful to save human life. Computer vision plays an important role in skin cancer detection. It has been proved its importance in detecting the cancer in its early stage which can be helpful to cure it. Accurate segmentation is one of the key steps in medical image diagnosis. Moreover, developing a precise segmentation of skin cancer images leads to better feature diagnosis, extractions, and classification. This paper develops a novel segmentation method for skin cancer images based on hybrid cross entropy thresholding techniques to find an optimum extraction of region that reflect the presence of skin cancer. The proposed methodology tackles the problem of finding the optimal thresholding using hybrid combination of both Gaussian and Gamma distributions. To evaluate the effectiveness of the proposed method, two benchmark skin lesion dermoscopic images datasets are: PH2 and ISIC 2017. The obtained results indicate that the proposed hybrid combination methodology gave better result and achieves 75% accuracy in skin cancer detection compared to other benchmark segmentation techniques.