Centre for Biomedical Engineering, Indian Institute of Technology Delhi, New Delhi 110016, India.
IEEE Trans Biomed Eng. 2010 Jun;57(6):1430-6. doi: 10.1109/TBME.2010.2041000. Epub 2010 Feb 17.
This paper proposes a novel and probably the first method, using Attanassov intuitionistic fuzzy set theory to segment blood vessels and also the blood cells in pathological images. This type of segmentation is very important in detecting different types of human diseases, e.g., an increase in the number of vessels may lead to cancer in prostates, mammary, etc. The medical images are not properly illuminated, and segmentation in that case becomes very difficult. A novel image segmentation approach using intuitionistic fuzzy set theory and a new membership function is proposed using restricted equivalence function from automorphisms, for finding the membership values of the pixels of the image. An intuitionistic fuzzy image is constructed using Sugeno type intuitionistic fuzzy generator. Local thresholding is applied to threshold medical images. The results showed a much better performance on poor contrast medical images, where almost all the blood vessels and blood cells are visible properly. There are several fuzzy and intuitionistic fuzzy thresholding methods, but these methods are not related to the medical images. To make a comparison with the proposed method with other thresholding methods, the method is compared with six nonfuzzy, fuzzy, and intuitionistic fuzzy methods.
本文提出了一种新颖的、可能是首例的方法，使用 Attanassov 直觉模糊集理论对病理图像中的血管和血细胞进行分割。这种分割在检测不同类型的人类疾病方面非常重要，例如，血管数量的增加可能导致前列腺、乳腺等部位的癌症。医学图像的照明不充分，在这种情况下，分割变得非常困难。本文提出了一种使用直觉模糊集理论和新隶属度函数的新图像分割方法，该方法使用自同构的受限等价函数来确定图像像素的隶属度值。使用 Sugeno 型直觉模糊生成器构建直觉模糊图像。对医学图像进行局部阈值处理。结果表明，在对比度差的医学图像上，该方法的性能要好得多，几乎可以清晰地显示出所有的血管和血细胞。有几种模糊和直觉模糊阈值方法，但这些方法与医学图像无关。为了将所提出的方法与其他阈值方法进行比较，将该方法与六种非模糊、模糊和直觉模糊方法进行了比较。
