• SKIN CANCER EXPERT SYSTEM USING FRACTAL DIMENSION

Abstract

There are many characteristics that differentiate normal moles (nevi) from melanomas. Fractal measures like fractal dimension and lacunarity are widely-used for the analysis of textures, or any images with self-similar content. However, all the existing approaches are defined for one dimensional signals or binary images with extension to grey-scale images. We proposed a novel approach, in combination to the probabilistic algorithm for the computation of the fractal dimension and lacunarity. Our method was used the boundary irregularity, which can be quantified using fractal dimension. In this work, fractal dimension of normal moles and melanoma was computed using the box counting method. We used this approach for the evaluation of the skin lesions, in case of psoriasis. The efficacy of these new descriptors was tested on a data set of 800 skin lesions, composed by 85 melanomas and nevi. The result showed the high accuracy of the developed GUI for melanoma detection.

Keywords- melanoma; skin cancer; fractal; cancer detection; fractal dimension.