Affiliations: [a]
School of Information and Communication Technology, Griffith University, Australia
| [b]
College of Science and Engineering, Texas A&M University-Corpus Christi, USA
Correspondence:
[*]
Corresponding author. Maedeh Beheshti, School of Information and Communication Technology, Griffith University, Australia. E-mail: [email protected].
Abstract: Accurate segmentation of fluorescence images has become increasingly important for recognizing cell nucleus that have the phenotype of interest in biomedical applications. In this study an ensemble based method is proposed for the segmentation of cell cancer microscopy images. The ensemble is constructed and compared using Bayes graph-cut algorithm, binary graph-cut algorithm, spatial fuzzy C-means, and fuzzy level set algorithm, which were chosen for their accuracy and efficiency in the segmentation area. We investigate the performance of each method separately and finally compare the results with the ensemble method. Experiments are conducted over two datasets with different cell types. At 95% confidence level, the ensemble based method represents the best among all the implemented algorithms. Also ensemble method depicts better results in comparison with other state-of-the-art segmentation methods.
