Automatic Recognition of Skin Cancer using Fully Convolution Networks and Conditional Random Fields
M. Bethu Dharani1, D. Venkatesan2

1M.Bethu Dharani, School of Computing, SASTRA Deemed University, Thanjavur (Tamil Nadu), India.
2D. Venkatesan, School of Computing, SASTRA Deemed University, Thanjavur (Tamil Nadu), India.

Manuscript received on 01 May 2019 | Revised Manuscript received on 15 May 2019 | Manuscript published on 30 May 2019 | PP: 441-445 | Volume-8 Issue-7, May 2019 | Retrieval Number: G5430058719/19©BEIESP
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 (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Skin cancer is one of the deadliest diseases that have been increasing all over the world. Automatic identification of lesion from low contrast dermoscopic images, over-segmentation of images and under-segmentation of images is a challenging task in the medical field. In order to overcome these challenges, we have proposed a Computerized Diagnosis system with deep fully convolution network (10*1 layer network) for segmenting the skin lesion which has been trained on end to end with 50% of dataset. Furthermore, Conditional Radom Field has been integrated with the existing framework for enhancing the segmentation performance and we added ensemble classifier technique called Bagging for accurate classification of lesion images into various categories. The proposed architecture is extensively evaluated on PH2 dataset. Experimental results showed that proposed method out performs well in comparison with the existing method. These results prove that the proposed system is more effective and suitable for any kind of medical images.
Keyword: Bagging, CAD, Conditional Random Field, Dermoscopy
Scope of the Article: Social Networks.