Figure 2.

Comparison of the segmentation quality between the developed tool and CAIMAN application. 1st row) original images: A and B are images from our database and C is an example image provided by CAIMAN with a less well-lit caption. 2nd row) segmentation performed by CAIMAN (green line) which does not detect small or very large vessels (D) and sometimes divides large vessels into small ones or considers blue areas as vessels (E). The image is well segmented (F). 3rdrow) result of the HSV segmentation (black line).In our images the morphometric tool makes a specific and good detection of the brown staining (G and H). The fine contour lines of the CAIMAN image have not been detected (I).4throw) a result of the radial analysis. The opened contour vessels have been properly closed, including the large vessel (J). The tool detected the absence of opened vessels in the second image (K). Finally, as it does not detect the fine contours in the CAIMAN image, it does not detect the opened vessels (L).

Fern├índez-Carrobles et al. Diagnostic Pathology 2013 8(Suppl 1):S20   doi:10.1186/1746-1596-8-S1-S20