Kaposi Sarcoma Case Study

3492 Words14 Pages
1 AUTOMATIC KAPOSI SARCOMA DETECTION USING TEXTURE DISTINTIVENESS Mrs.S.Haseena, Assistant Professor, Department of IT, Mepco Schlenk Engineering College, Sivakasi. Tamilnadu,India. haseena@mepcoeng.ac.in, Abstract— As there is a growing emphasis on skin cancer detection, Kaposi sarcoma has recently received increasing attention. Kaposi sarcoma is one deadliest form of skin cancer. The time and costs required for dermatologists to screen all patients for Kaposi sarcoma are prohibitively expensive. There is a need for an automated system to assess a patient's risk of Kaposi sarcoma using photographs of their skin lesions. Dermatologists could perform diagnosis without the need of special or expensive equipment. One challenge in implementing…show more content…
Recent work with automated Kaposi sarcoma screening algorithms tries to adapt the algorithms to analyze images taken by a standard digital camera. The skin lesion image taken by standard digital camera shown in Figure1. Figure 2 shows the flow chart for an automated skin lesion diagnosis. Most segmentation algorithms for dermatological images or photographs use color information, either in a single channel or across three color channels, to find the lesion. Another approach to find skin lesions is to incorporate textural information, because normal skin and lesion areas have different textures. Textures include smoothness, roughness, or2 the presence of ridges, bumps or other deformations and are visible by variation in pixel intensities in an area. Features and measurements of a texture in an image are extracted and textures from different regions are compared. Stoecker[7] analyzed texture in skin images using basic statistical, such as the gray-level cooccurrence matrix. They found that texture analysis could accurately find regions with a smooth texture and that texture analysis is applicable to segmentation and classification of dermatological…show more content…
 K ,  1 ,  2 ,......  K ,  1 ,  2 ,......  K  Where, μ is an distribution mean and  is a distribution covariance. Here, there is No closed form solution exists for5 Eqn.8 in general, so an expectation-maximization iterative algorithm is used. The expectation-maximization algorithm is initialized using cluster means, covariance and mixing proportions based on the results of the k-means clustering. Expectation-maximization is an iterative algorithm. The initial parameters for the Gaussian mixture model are obtained from the results of the K-means clustering. That is, the initial Gaussian means are equal to the k-means cluster means as mentioned in [1] and the distribution covariance and mixing proportions are also dependent on the cluster results. μ = μ (10) = (11) Figure 11: Learning representative texture distribution V. EXPERIMENTAL RESULTS In this section, we explains comparison of the proposed TDLS algorithm and Otsu-RGB segmentation algorithm. The Otsu segmentation technique is tested on simple RGB skin lesion image. Figure 12 and 13 shows the results perform based on TDLS and Otsu-RGB segmentation algorithm. Image Otsu-RGB TDLS Figure 12: Experimental results

    More about Kaposi Sarcoma Case Study

      Open Document