Skin Cancer Diagnostic using Machine Learning Techniques – Shearlet Transform and Naïve Bayes Classifier
S. Mohan Kumar1, J. Ram Kumar2, K. Gopalakrishnan3
1Dr. S. Mohan Kumar, Department of Mechanical Engineering, Indian Institute of Technology, Kanpur, India.
2Dr. J. Ram Kumar, Department of Mechanical Engineering, Indian Institute of Technology, Kanpur, India.
3Dr. K. Gopalakrishnan, New Horizon College of Engineering, Bangalore, Karnataka, India.
Manuscript received on November 21, 2019. | Revised Manuscript received on December 15, 2019. | Manuscript published on December 30, 2019. | PP: 3478-3480 | Volume-9 Issue-2, December, 2019. | Retrieval Number: B4916129219/2019©BEIESP | DOI: 10.35940/ijeat.B4916.129219
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Development of abnormal cells in the skin is known as skin cancer or melanoma, which can spread other parts of the body. Melanoma rarely occurs in eye, mouth and intestines. In this study, the classification of melanoma using shearlet transform coefficients and naïve Bayes classifier is discussed. The melanoma images are decomposed by the shearlet transform. Then, from the shearlet coefficients, predefined number of (50, 75 and 100) coefficients are selected from the decomposed subbands. The selected subband coefficients are directly applied to the naïve Bayes classifier. Performance of skin cancer classification system is measured in terms of accuracy. Results show that a better classification accuracy of 90.5 % is achieved at 3rd level with 100 coefficients of shearlet transform and naïve Bayes classifier for skin image classification system.
Keywords: Melanoma, Shearlet transform, Subband coefficients, Naïve Bayes classifier.