Benchmark images

This web site provides the benchmark images presented in article Benchmark set of synthetic images for validating cell image analysis algorithms (preprint of an article published in the Proceedings of the 16th European Signal Processing Conference (EUSIPCO-2008) in 2008, published by EURASIP). The synthetic images and corresponding ground truth images, along with the options file used for generating the images can befound below.

Description of data: Each compressed package (.zip) contains a set of images (.tif). Filenames start with numbers, for each number there is a pair of images. Filename ending RGB refers to the simulated image, whereas BW denotes the validation image, where each channel represents the binary ground truth or the corresponding channel in RGB image.

The SIMCEP options files (.m) for generating the database images are provided. Thus, more data with similar characteristics can be generated. Also further experimenting with other parameter values is encouraged.

Set 1: Clustering with increasing probability

Images Objects/
image
Probability of
clustering
Overlap Options file
Download (21,5MB) 300 p = 0 L = 0 Download
Download (21,2MB) 300 p = 0.15 L = 1 Download
Download (21,3MB) 300 p = 0.30 L = 1 Download
Download (21,0MB) 300 p = 0.45 L = 1 Download
Download (20,8MB) 300 p = 0.60 L = 1 Download

Set 2: Cells with nuclei, cytoplasm and subcellular objects

Images Objects/
image
Subcell. obj./
cell
Probability of
clustering
Overlap Options file
Download (26,9MB) 50 * 3 p = 0 L = 0 Download
Download (38,0MB) 50 * 3 p = 0.25 L = 1 Download

* Note that the number of objects per image is set to 50 instead of 40 mentioned in the article.

Set 3: Cells from two populations

Images Objects/
image
Radius Shape parameters
(alpha and beta)
Options file
Download (19,3MB) 100 12 0.1 Download
Download (18,9MB) 100 11 0.2 Download

Page created
22.02.2008
Last modified
17.05.2009