中文摘要：

高效准确地检测细胞核是自动分析病理学组织的重要部分,也是细胞核准确分割的第一步,分割的准确性很大程度上取决于种子点检测的准确性与可靠性。本文提出一种改进的投票算法来检测细胞核种子点。首先,通过对乳腺病理切片图像进行预处理,利用Otsu阈值方法对预处理后的图像进行粗分割;然后用椭圆拟合方法检测出部分细胞核种子点;最后利用改进的投票算法检测细胞核种子点。实验结果表明,本文所提出的算法可以准确地将病理切片图像中的细胞核检测出来,且检测精确度超过90%,能够提供比现有方法更好的检测性能,可用于定量分析乳腺病理图像。

英文摘要：

Efficient and accurate detection of nuclei is an important step in histopathology for disease diagnosis, and the first step for nuclear segmentation. The accuracy of segmentation depends critically on the accuracy and reliability of the seed point detection. Herein, an improved voting algorithm is proposed to detect nuclear seed points. Firstly, Otsu threshold method is used to segment the preprocessed pathological section of breast. And then, some nuclear seed points were detected with ellipse fitting method. Finally, the improved voting algorithm is used to detect nuclear seed points. The experimental results show that the improved voting algorithm can accurately detect the nuclear seed points in the pathological section and the detection accuracy is more than 90%, which means the proposed algorithm can provide better detection performance than the existing methods. The improved voting algorithm can be used for the quantitative analysis of breast pathological section.