Parallel Processing Strategies for Cell Motility and Shape Analysis - High-Throughput Image Reconstruction and Analysis - page 52

Figure 3.6 Cell-to-cell interaction cases: (a) case 1: separate cells, one-to-one temporal overlap;
(b) case 2: merged cells one-to-one temporal overlap; (c) case 3: spatially separate and temporally
merged; and (d) case 4: spatially merged and temporally merged.
3.2.2 Edge-Based Active Contour Cell Segmentation
While region-based active contours have been succesfully used in many cell and nu-
cleus segmentation applications [1--3, 24], they require a bimodal image model to
discern background and foreground which is not applicable in some high-resolution
cell images. Figure 3.8 shows typical high-resolution phase contrast images of cells,
where white regions surrounding the cells are the phase halos. A straightforward
implementation of [24] converges to the boundaries between phase halo and the
rest of the image. For images where intensity inside the cell boundaries is highly
heterogeneous, and there is no clear difference between inside and outside inten-
sity profiles, edge-based approaches are used since region-based active contour
approaches are not applicable.
In this section, we describe two edge-based level set segmentation methods
and their performance for handling phase halos. The first technique is the classi-
cal geodesic active contour approach [19]; the second technique is our extension
directed edge profile-guided level set active contours. Compared to region-based
Figure 3.7 Cell-to-cell interaction case 2 test sequence frames #1, 8, 13, 20. Row 1: four color
level set segmentation; Row 2: tracking results.