Dermoscopic Feature Analysis for Melanoma Recognition and Prevention Uzma Jamil Member IEEE, Ph.D Scholar, Department of Computer Engineering, Bahria University, Islamabad, Pakistan. uzma_gcuf@yahoo.com Dr. Shehzad Khalid Professor, Head, Department of Computer Engineering, Bahria University, Islamabad, Pakistan. shehzad_khalid@hotmail.com Dr. M.Usman Akram Assistant Professor, National University of Sciences & Technology, Islamabad, Pakistan. usmakram@gmail.com Abstract—Computer-aided diagnosis system automatically analyze skin lesions , and reduces the amount of repetitive and boring tasks carried out by the doctor. Recognize the full model, which includes three important stages in order to comply with the lesion analysis : segmentation and feature extraction and classification systems. The data-set contains images and annotations provided by physicians. Segmentation is an imperative preprocessing step for CAD system of skin lesions. Feature extraction of segmented skin lesions is a pivotal step for implementing accurate decision support systems. Dermatologists take keen interest in examining a specific clinically significant part in a lesion. That part is projected to have lesion information in the form of texture that can be relevant for detection. In case of detection of melanoma various local features for example pigment network and streaks usually occur in peripheral region of the lesion. This led to the extraction of peripheral part for feature extraction instead of whole lesion processing. In this article detailed information regarding Feature extraction and selection techniques for dermoscopic images is presented. Index Terms— skin cancer, dermoscopy, features, image processing, pattern recognition. INTRODUCTION The mortality trends and statistics shows that skin cancer is the fourth deadly type of cancer worldwide. Usually, skin cancer is divided into two main types: melanoma and non-melanoma as depicted in Fig.1. Melanoma is among the deadliest skin cancer, rapidly increasing in the world. The American Cancer Society estimates that more than 135,000 new melanoma cases are diagnosed in the United States . In 2015, an estimated 73,870 new cases are diagnosed of Melanoma, and about 42,670 men , women 31200 [1]. Vital warning symptoms and signs of melanoma includes mole that is varying in its size, shape, or color or the appearance of a new growth of the mole on the skin. Along with Melanoma, Non-melanoma skin cancers (NMSC) consists of basal 1 and squamous cell carcinomas 2 . NMSC are 1 Basal cell carcinoma (BCC) are the lesions that arise in the skin’s basal layer, the deepest layer of the epidermis. 2 Squamous cell carcinoma (SCC) are the lesion that arise in the squamous cells, which is the upper layer of the epidermis also very difficult to diagnose at its early stage because of similarity in appearance with normal skin [3]. The estimated occurrence of NMSC cases in 2015 is about 3.5 million cases. NMSC is highly curable if properly diagnosed at its initial stage. For an early detection of skin cancer the best way is to recognize new or changing skin growths, mainly those that look different from other moles. Fig.1. Skin Cancer Types (a) Squamous Cell Carcinoma, (b) Basal Cell Carcinoma and (c) Melanoma (Reprint from [2]) Dermoscopy is a non persistent method used for estimation of skin lesions. This process allows the specific , minute pigment pattern evaluation to improve differential diagnosis. Doctors apply gel to the skin , and magnification instrument ( dermoscopy , stereoscopic microscope or a digital imaging system ) amplifies the lesion and check it [2]. After amplification allows multiple surface and subsurface structure, which is not visible to the naked eye recognition [4]. Brown spots become more prominent features and a pigment pattern clearly with the help of the skin can be seen under the microscope.