Translator Disclaimer
26 April 2018 Modeling of skin cancer dermatoscopy images
Author Affiliations +
An early identified cancer is more likely to effective respond to treatment and has a less expensive treatment as well. Dermatoscopy is one of general diagnostic techniques for skin cancer early detection that allows us in vivo evaluation of colors and microstructures on skin lesions. Digital phantoms with known properties are required during new instrument developing to compare sample’s features with data from the instrument. An algorithm for image modeling of skin cancer is proposed in the paper. Steps of the algorithm include setting shape, texture generation, adding texture and normal skin background setting. The Gaussian represents the shape, and then the texture generation based on a fractal noise algorithm is responsible for spatial chromophores distributions, while the colormap applied to the values corresponds to spectral properties. Finally, a normal skin image simulated by mixed Monte Carlo method using a special online tool is added as a background. Varying of Asymmetry, Borders, Colors and Diameter settings is shown to be fully matched to the ABCD clinical recognition algorithm. The asymmetry is specified by setting different standard deviation values of Gaussian in different parts of image. The noise amplitude is increased to set the irregular borders score. Standard deviation is changed to determine size of the lesion. Colors are set by colormap changing. The algorithm for simulating different structural elements is required to match with others recognition algorithms.
© (2018) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Malica B. Iralieva, Oleg O. Myakinin, Ivan A. Bratchenko, and Valery P. Zakharov "Modeling of skin cancer dermatoscopy images", Proc. SPIE 10716, Saratov Fall Meeting 2017: Optical Technologies in Biophysics and Medicine XIX, 107160C (26 April 2018);

Back to Top