Lee Hyunju, Kwon Kiwoon
Department of Mathematics, Dongguk University, Seoul, Seoul 04620, Korea.
Phys Med Biol. 2017 Mar 7;62(5):1865-1884. doi: 10.1088/1361-6560/aa562f. Epub 2016 Dec 30.
The medical community currently employs the ABCD (asymmetry, border irregularity, color variegation, and diameter of the lesion) criteria in the early diagnosis of a malignant melanoma. Although many image segmentation and classification methods are used to analyze the ABCD criteria, it is rare to see a study containing mathematical justification of the parameters that are used to quantify the ABCD criteria. In this paper, we suggest new parameters to assess asymmetry, border irregularity, and color variegation, and explain the mathematical meaning of the parameters. The suggested parameters are then tested with 24 skin samples. The parameters suggested for the 24 skin samples are displayed in three-dimensional coordinates and are compared to those presented in other studies (Ercal et al 1994 IEEE Trans. Biomed. Eng. 41 837-45, Cheerla and Frazier 2014 Int. J. Innovative Res. Sci., Eng. Technol. 3 9164-83) in terms of Pearson correlation coefficient and classification accuracy in determining the malignancy of the lesions.
医学界目前采用ABCD（不对称性、边界不规则性、颜色斑驳度和病变直径）标准来早期诊断恶性黑色素瘤。尽管使用了许多图像分割和分类方法来分析ABCD标准，但很少见到有研究对用于量化ABCD标准的参数进行数学论证。在本文中，我们提出了用于评估不对称性、边界不规则性和颜色斑驳度的新参数，并解释了这些参数的数学意义。然后，我们用24个皮肤样本对所提出的参数进行了测试。针对24个皮肤样本所提出的参数以三维坐标形式呈现，并在确定病变恶性程度方面，就皮尔逊相关系数和分类准确率与其他研究（Ercal等人，1994年，《IEEE生物医学工程汇刊》41卷，837 - 45页；Cheerla和Frazier，2014年，《国际创新研究科学、工程与技术杂志》3卷，9164 - 83页）中呈现的参数进行了比较。
