Combining Image Thresholding and Fast Marching for Nuclei Extraction in Microscopic Images
Computer-Aided Diagnosis (CAD) in digital pathology very often boils down to examination of nuclei using morphological analysis. To determine the characteristics of nuclei, they need to be segmented from the background or other objects in the image (e.g. red blood cells). Despite a tremendous work that has been done to improve segmentation methods, nuclei segmentation remains a very challenging problem. This particularly applies to the cytological images, where nuclei often touch, overlap, cluster, are obscured, or destroyed. Most well known methods of image processing cannot cope with this challenge. Nevertheless, in this study we demonstrated that methods like image thresholding, edge detection, erosion and fast marching, when combined, give satisfactory segmentation results. The proposed approach uses isodata image thresholding and Canny edge detection to find nuclei regions in the image. Then this information is employed to determine centers of the nuclei using conditional erosion. Finally, fast marching algorithm extracts nuclei. The method was applied to extract nuclei from microscopic images of cytological material obtained from breast. Different morphometric, textural, colorimetric and topological features were computed for segmented nuclei to describe cases (patients). The effectiveness of the segmentation was evaluated in terms of classification accuracy of breast cancer, where the cases were classified as either benign or malignant. The acquired predictive accuracy was 98 %, which is very promising and shows that the presented method ensures accurate nuclei segmentaion in cytological images.
KeywordsNuclei segmentation Image thresholding Fast marching Breast cancer Cytology
The research was supported by National Science Centre, Poland (2015/17/B/ST7/03704).
- 3.Jeleń, L., Fevens, T., Krzyżak, A.: Classification of breast cancer malignancy using cytological images of fine needle aspiration biopsies. Int. J. Appl. Math. Comp. Sci. 18(1), 75–83 (2010)Google Scholar
- 6.Kruk, M., Osowski, S., Markiewicz, T., Slodkowska, J., Koktysz, R., Kozlowski, W., Swiderski, B.: Computer approach to recognition of fuhrman grade of cells in clear-cell renal cell carcinoma. Anal. Quant. Cytopathol. Histpathol. 36(3), 147–160 (2014)Google Scholar
- 8.Ridler, T.W., Calvard, S.: Picture thresholding using an iterative selection method. IEEE Trans. Syst. Man Cybern. SMC–8(8), 630–632 (1978)Google Scholar