Fast Globally Optimal Segmentation of Cells in Fluorescence Microscopy Images

Abstract

Abstract. Accurate and efficient segmentation of cells in fluorescence microscopy images is of central importance for the quantification of protein expression in high-throughput screening applications. We propose a new approach for segmenting cell nuclei which is based on active contours and convex energy functionals. Compared to previous work, our approach determines the global solution. Thus, the approach does not suffer from local minima and the segmentation result does not depend on the initialization. We also suggest a numeric approach for efficiently computing the solution. The performance of our approach has been evaluated using fluorescence microscopy images of different cell types. We have also performed a quantitative comparison with previous segmentation approaches.

Unsupervised cell identification on multidimensional X-ray fluorescence datasets

“…Therefore, it is necessary to model how different cells overlap with each other. Nevertheless, most known methods do not model cell overlap explicitly; instead they either generate nonoverlapping regions from overlapping cells (Arteta et al, 2012;Lin et al, 2003) or only process cells at most touching at their boundaries (Bergeest & Rohr, 2011. In x7, we will discuss the difference between our method and a representative cell counting routine ('analyze particle' from ImageJ) in more detail.…”

Cell nuclei segmentation in fluorescence microscopy images using inter- and intra-region discriminative information

“…To demonstrate the effectiveness of our method, we have compared with four other methods: (1) Otsu thresholding; (2) watershed; (3) level set with handling of intensity inhomogeneity [18] with watershed outputs as the initial contours; and (4) the state-of-the-art [2] reported for the same datasets. For the convenience of comparison, the same performance metrics used in [2] are used here, including the Dice coefficient for pixel-level labeling, the normalized sum of distances (NSD) and Hausdorff distance (HD) for contour delineation, and the number of false positives (FP) and false negatives (FN) for object-level detection.…”

“…For the convenience of comparison, the same performance metrics used in [2] are used here, including the Dice coefficient for pixel-level labeling, the normalized sum of distances (NSD) and Hausdorff distance (HD) for contour delineation, and the number of false positives (FP) and false negatives (FN) for object-level detection. Except Dice, the other four measures are the lower the better.…”

“…Song background. More advanced methods, especially those based on level sets [2], [3], are also becoming a major trend for boundary delineation with implicit contour modeling. Another category of methods is focused on pixel-wise labeling accuracy, with various classification algorithms [4], [5], [6] or graphical models [7], [8], [9], [10].…”

Efficient globally optimal segmentation of cells in fluorescence microscopy images using level sets and convex energy functionals

“…Based on the convex functionals, we developed a three-step approach and a two-step approach for cell nucleus segmentation. The three-step approach (Bergeest and Rohr, 2011) combines the convex region-based Chan-Vese functional and the convex Bayesian functional, and can cope with global intensity inhomogeneities. In this contribution, we also introduce a new two-step approach which combines the convex Bayesian functional and the convex region-scalable fitting energy functional.…”