Fence size and live-cell versus non-cell objects. Individual track segments were manually scored from image sequences according to whether they originated from live-cell or non-cell objects. A custom viewing device permitted selection of a maximum fence size for tracks to be included in the display; in addition, only segments from tracks connecting 4 or more points were displayed. With each increasing step in fence size, the entire image sequence consisting of more than 100 images was re-scored. This process was repeated for 4 representative wells for each cell type. The cumulative total counts for live-cell (blue) and non-cell (pink) objects are plotted versus fence size on the horizontal axis for both cell types, KM101 (left) and MG-63 (right). Cell debris appeared to be more abundant from MG-63 cells after plating, but the wider range of permissible area for MG-63 cells would be expected to include also more non-cell objects of larger area. Objects of larger area would be expected to have greater segmentation and centroid position variability (see Segmentation/Outline Error) resulting in greater associated fence size. For MG-63 cells, the counting of non-cell objects was discontinued at fence sizes of 40 and above because it became simply too demanding to distinguish and count both live and non-cell objects.
Bahnson et al. BMC Cell Biology 2005 6:19 doi:10.1186/1471-2121-6-19