Nuclei enhancement and segmentation in color cervical smear images

• Published in: 2014 IEEE International Conference on Robotics and Biomimetics (ROBIO 2014) pp. 107 - 112
• Main Authors: Tao Guan, Dongxiang Zhou, Weihong Fan, Keju Peng, Chao Xu, Xuanping Cai
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-12-2014
• Subjects: Adaptive equalizers; Feature extraction; Gray-scale; Histograms; Image color analysis; Image reconstruction; Image segmentation; Microscopic image processing; morphological reconstruction; nuclei enhancement; nuclei segmentation
• DOI: 10.1109/ROBIO.2014.7090315

Cell image segmentation is one of the hot topics in medical image processing. Most of the classical cell image segmentation algorithms perform the segmentation directly on the original image and result in the loss of the cell nuclei with low intensity contrast. To solve this problem, this paper presents a novel nuclei segmentation method. Based on analyzing the characteristics of the cell nuclei, we first enhance the nuclei according to their unique features in the image. The proposed nuclei enhancement method combines the intensity and the color information of the image, and is thus effective to enhance the nuclei with relatively low intensity contrast. Then, the morphological reconstruction is employed to extract the regional maxima of the enhanced image, and several shape description parameters are finally used to screen out the true cell nuclei from the extracted regions. Experiments have been performed on real cervical smear images, and the results validate the effectiveness of the proposed method for nuclei segmentation in cervical smear images.