Accession number BBBC007 · Version 1
Drosophila melanogaster Kc167 cells were stained for DNA (to label nuclei) and actin (a cytoskeletal protein, to show the cell body). Automatic cytometry requires that cells be segmented, i.e., that the pixels belonging to each cell be identified. Because segmenting nuclei and distinguishing foreground from background is comparatively easy for these images, the focus here is on finding the boundaries between adjacent cells.
Images were acquired using a motorized Zeiss Axioplan 2 and a Axiocam MRm camera, and are provided courtesy of the laboratory of David Sabatini at the Whitehead Institute for Biomedical Research. Each image is roughly 512 x 512 pixels, with cells roughly 25 pixels in dimeter, and 80 cells per image on average. The two channels (DNA and actin) of each image are stored in separate gray-scale 8-bit TIFF files.
First, nuclei were outlined by hand. The nuclear outlines were overlaid on the cell images, and one cell per nucleus was outlined. Outlines of nuclei and cells are stored in separate monochrome 8-bit TIFF files.
Algorithms for finding boundaries between adjacent cells may use the union of the manually outlined cells as input. They may also use the manually outlined nuclei as seed regions. This ensures that the algorithms are compared according to their ability to find boundaries betwen adjacent cells, and not the orthogonal problems of segmenting nuclei and distinguishing foreground (cells) from background.
To compare an algorithm's results to the manual outlines, define the relevant boundary pixels as the pixels that are on the boundary found by the algorithm and that are not adjacent to any background pixels. For each relevant pixel, compute the Euclidean distance to the corresponding pixel on the manually found outline. Report the percentage of relevant pixels that are within two pixels of the corresponding pixel on the manually found outline.
Published results using this image set
|64%||Jones et al., in the Proceedings of the ICCV Workshop on Computer Vision for Biomedical Image Applications (CVBIA), 2005.|
We would like to thank Chris Gang for outlining nuclei and cells.
"We used image set BBBC007v1 image set version 1 (Jones et al., Proc. ICCV Workshop on Computer Vision for Biomedical Image Applications, 2005), available from the Broad Bioimage Benchmark Collection [Ljosa et al., Nature Methods, 2012]."
To the extent possible under law, Anne Carpenter has waived all copyright and related or neighboring rights to Drosophila Kc167 images and ground truth. This work is published from: United States.