What it is
Simulated high-content screening images of fluorescent cell populations, generated with the SIMCEP platform. Focus blur was simulated with Gaussian filters, so the same population appears from sharp to very blurred. Because the images are synthetic, the true number of cells in every image is known.
| Fact | Value |
|---|---|
| Accession | BBBC005, version 1 |
| Images | 19,200 files (zip, 1.8 GB) |
| Format | 8-bit TIF, 696 x 520 px |
| Cells per image | 1 to 100 |
| Blur levels | 1 to 48 |
| Stains | stain 1 (cell body) or stain 2 (nuclei) |
| Layout | 384-well plate naming (rows A to P, columns 1 to 24) |
| License | CC0 1.0 |
Ground truth
- Counts for every image, encoded in the file name.
- Foreground masks for the images that are fully in focus (file names with "F1"): 1,200 8-bit TIF files, white foreground on black.
- A published nucleus-count results file (Bray et al., 2011) for 9,600 fields of view. The source page notes this is a results file, not ground truth.
What people use it for
Counting cells without segmenting them, testing how a segmentation method degrades with blur, and choosing a focus metric. Clustering was simulated at 25%, so touching cells are part of the test.
Source and citation
Source: https://bbbc.broadinstitute.org/BBBC005 (read 2026-10-22). Cite: "We used the image set BBBC005v1 from the Broad Bioimage Benchmark Collection [Ljosa et al., Nature Methods, 2012]."
Annotate your own microscopy images
Count, outline and review cells with your team in the browser.
Frequently asked questions
How many images are in BBBC005?
19,200 image files, 8-bit TIF, 696 x 520 pixels.
Where is the cell count for each BBBC005 image?
It is encoded in the file name. The 1,200 foreground masks for the in-focus images carry the count in their file names too.
What is the license of BBBC005?
CC0 1.0. Anne Carpenter waived all copyright and related rights.
