一种低分辨率细胞显微图像的分割与统计

Segmentation and Statistics of Low Resolution Cell Microscopic Images

针对医学显微图像的低清晰度与染色剂干扰导致细胞识别不准确的问题,本文提出了一种基于K均值(K-means)聚类与Canny算子分割相结合的方法用于实现细胞图像的自动识别与统计。K-means聚类法作为主算法从背景中分割出绝大多数染色正常的细胞,Canny算子作为补充方法用于分割染色度不足的细胞,或者是与染色污染块粘连的细胞。具体做法是,先用K-means聚类法分割细胞图像,获取包含细胞核、细胞质和环境背景的三值图像,通过提取细胞核确定每个细胞的中心位置;而对于其它因染色问题,用K-means法未能正常分割的细胞,采用Canny算子获取该细胞边界,补充遗漏的细胞的统计。最后,算法利用细胞的特征参数进一步提高识别的准确性。实验结果表明,该算法降低了对图像质量的要求,能够自动排除染色剂的干扰,快速有效地识别有核质差异的细胞图像。算法统计精度达到了97%以上。

Aiming at the problem of inaccurate cell recognition caused by low definition of medical microscopic images and interference of dyes,this paper proposed a method based on the combination of K-means clustering and Canny operator segmentation to realize automatic recognition and statistics of cell images.K-means clustering was used to segment the majority of normal stained cells from the background.Canny operator was used as a supplementary method to segment the cells with insufficient staining degree,or the cells adhered to the staining contaminated mass.The cell image is firstly segmented by K-means clustering method to obtain the three-value image.For other cells that could not be normally segmented by the K-means method due to dyeing problems,Canny operator was used to obtain the boundary of the cell to supplement the statistics of the missing cells.Finally,the algorithm made use of the characteristic parameters of cells to further improve the accuracy of recognition.The experimental results show that the algorithm can reduce the requirement of image quality,automatically exclude the interference of chromatids,and quickly and effectively recognize the cell images with nucleo-cytoplasmic differences.The statistical accuracy of the algorithm is over 97%.