Automatic Detection and Analysis of Melanoma Skin Cancer using Dermoscopy Images
Mahmudul Hasan1, Mohammad Mohsin1, Md. Kamal Hossain Chowdhury1

1Mahmudul Hasan, Asst. Prof., Dept. of CSE, Comilla University, Cumilla, Bangladesh.
2Mohammad Mohsin, Dept. of CSE, Comilla University, Cumilla, Bangladesh.
3Md Kamal Hossain Chowdhury, Asst. Prof., Dept. of CSE, Comilla University,Cumilla, Bangladesh. 

Manuscript received on 1 August 2019. | Revised Manuscript received on 10 August 2019. | Manuscript published on 30 September 2019. | PP: 2116-2122 | Volume-8 Issue-3 September 2019 | Retrieval Number: C4561098319/19©BEIESP | DOI: 10.35940/ijrte.C4561.098319
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: Skin cancer is known as one of the most risky types of cancer. Several kinds of skin cancer, such as melanoma, basal and squamous cell carcinoma, etc., are available. The most unpredictable cancer is melanoma. If we can detect melanoma skin cancer at an early stage, the chances of recovery will be good and we can save many valuable lives. But if we fail to detect early, melanoma can disperse to the different parts of the body and chance of recovery will become difficult. This research presents a developed system to do melanoma diagnosis by using several dermoscopy images. In this research, we preprocessed the images to remove hairs and noises by using some filter techniques such as dull razor technique, median filtering, etc. After that, we segmented the image to find the infected area using some segmentation method and we choose the method that will give us the best results. Then we post-process the images and choose the most infected lesion. After segmentation of the skin lesion, we checked the segmentation accuracy concerning some basic criteria. We compared the segmented skin lesions with the marked skin lesions by a dermatologist. Then we extracted the features of the images of different criteria, such as Asymmetry, Border irregularity, Color variance, Diameter which have the acronym as ABCD. We also analyzed the texture of the lesions and extracted the geometrical features. Finally, we choose decision tree classification methods that gave us the best results.
Keywords: Feature extraction, Image segmentation, Object recognition, Pattern clustering, Pattern matching .

Scope of the Article:
Predictive Analysis