Melanoma Segmentation and Classification using Deep Learning
R. D. Seeja1, A. Suresh2

1R. D. Seeja* , Research Scholar, Department of Computer Science, Periyar University, Salem, Tamil Nadu, India.
2Dr. A Suresh, Principal, Siri PSG Arts and Science College for women, Sankagiri, Salem-637301.,Tamil Nadu, India.

Manuscript received on September 16, 2019. | Revised Manuscript received on 24 September, 2019. | Manuscript published on October 10, 2019. | PP: 2667-2672 | Volume-8 Issue-12, October 2019. | Retrieval Number: L25161081219/2019©BEIESP | DOI: 10.35940/ijitee.L2516.1081219
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (

Abstract: Melanoma is the most destructive form of skin cancer. Early diagnosis of melanoma can be curable. At the same time accurate diagnosis is very essential because of the similarities of melanoma and benign lesions. Hence computerized recognition approaches are highly demanded for dermoscopy images. The main purpose of this research is to develop an automatic system to improve the classification performance of melanoma.The effectiveness of this framework is evaluated on ISBI 2016 Skin Lesion Analysis towards Melanoma Detection Challenge dataset. Initially deep learning based U-Net algorithm is used to segment the lesion region from the nearby healthy skin and then extract discriminate features with the help of Convolutional Neural Network. VGG16 Net algorithm is used to classify every lesion in a dermoscopic image as a Benign or Melanoma. Results are found from classification with and without segmented images. Classification with segmented images produces accuracy of 83.18%, Sensitivity of 95.53%, and specificity of 96.22%. Based on these values the deep learning based classification with segmented images produces better result and it helps to improve the diagnosis performance. The proposed method would constitute a valuable support for physicians in every day clinical practice.
Keywords: Melanoma, Deep learning, Dermoscopy, Convolutional Neural Networks, Lesion Segmentation.
Scope of the Article: Deep learning