一种新的组织病理图像阳性细胞轮廓检测方法

组织病理图像中阳性细胞比例的检测对癌症和肿瘤的定性和定级起决定作用。提出一种用于细胞准确计数的新的轮廓检测方法,针对组织病理图像色彩纹理复杂、细胞边界模糊等特点,结合通道提取和图像二值化方法实现阳性细胞的准确分离,并在CV模型基础上完成对细胞的轮廓提取。实验表明,该方法有效解决了传统方法无法处理的弱边缘问题,在保持算法性能的前提下,可自动分离组织病理图像中的阳性细胞并检测其轮廓。

a New Algorithm of Positive Cells Contour Detection in Histopathology Image

The ratio of positive cells in histopathology image affects definitively diagnosis and classification of cancer. In allusion to complex color and texture, fuzzy boundary in histopathology image, this paper proposed a new algorithm of positive cells contour detection for cells counting, which combined RGB channel extracting and image binaryzation to separate positive cells, and have high efficiency in contour detection of these cells base on CV model. Experiment demonstrates that, the algorithm solves the problem of weak boundaries that can`t achieve in traditional way, and on the premise of maintaining the efficiency of algorithm, this method can separate the positive cells and detect its contour automatically.