A spectral k-means approach to bright-field cell image segmentation

• Published in: 2010 Annual International Conference of the IEEE Engineering in Medicine and Biology Vol. 2010; pp. 4748 - 4751
• Main Authors: Bradbury, L, Wan, J W L
• Format: Conference Proceeding Journal Article
• Language: English
• Published: United States IEEE 01.01.2010
• Subjects: Active contours; Algorithms; Animals; Approximation algorithms; Artificial Intelligence; Cell Line; Cell Tracking - methods; Clustering algorithms; Image edge detection; Image Enhancement - methods; Image Interpretation, Computer-Assisted - methods; Image segmentation; Mice; Microscopy; Muscle Fibers, Skeletal - cytology; Pattern Recognition, Automated - methods; Pixel; Reproducibility of Results; Sensitivity and Specificity; Subtraction Technique
• ISBN: 1424441234; 9781424441235
• ISSN: 1094-687X; 1557-170X
• DOI: 10.1109/IEMBS.2010.5626380

Automatic segmentation of bright-field cell images is important to cell biologists, but difficult to complete due to the complex nature of the cells in bright-field images (poor contrast, broken halo, missing boundaries). Standard approaches such as level set segmentation and active contours work well for fluorescent images where cells appear as round shape, but become less effective when optical artifacts such as halo exist in bright-field images. In this paper, we present a robust segmentation method which combines the spectral and k-means clustering techniques to locate cells in bright-field images. This approach models an image as a matrix graph and segment different regions of the image by computing the appropriate eigenvectors of the matrix graph and using the k-means algorithm. We illustrate the effectiveness of the method by segmentation results of C2C12 (muscle) cells in bright-field images.