Cell segmentation via spectral analysis on phase retardation features

In this paper, the authors propose a cell segmentation algorithm via spectral analysis over phase retardation features, which are derived from the optical principle of phase contrast microscopy image formation process. Images are first partitioned into phase-homogeneous atoms by clustering neighboring pixels with similar phase retardation features. Cell segmentation is then performed by clustering the atoms into several clusters using multi-class spectral analysis. Experimental results demonstrate that our method generates quality cell segmentation results and outperforms previous methods.