Segmentation of Individual Cells in Phase Contrast Microscopy Images

0435026 - ÚTIA 2015 RIV GB eng C - Konferenční příspěvek (zahraniční konf.)
Soukup, Jindřich - Lašan, M. - Šroubek, Filip
Segmentation of Individual Cells in Phase Contrast Microscopy Images.
Proceedings of the 18th Conference on Medical Image Understanding and Analysis. London: City University London, 2014 - (Reyes-Aldasoro, C.; Slabaugh, G.), s. 185-190. ISBN 1-901725-51-0.
[Medical Image Understanding and Analysis 2014. London (GB), 09.07.2014-11.07.2014]
Grant CEP: GA ČR GA13-29225S
Grant ostatní: GA MŠk(CZ) LO1205; GAJU(CZ) 134/2013/Z; GA UK(CZ) 914813/2013; OP VaVpI(CZ) CZ.1.05/2.1.00/ 01.0024
Institucionální podpora: RVO:67985556
Klíčová slova: image segmentation * phase contrast microscopy * time-lapse imaging
Kód oboru RIV: JD - Využití počítačů, robotika a její aplikace
http://library.utia.cas.cz/separaty/2014/ZOI/soukup-0435026.pdf

Phase contrast microscopy is widely used for live cell imaging. Automatic computer processing of phase-contrast images is complicated, particularly because of the halos that are present around the cells. We introduce a new algorithm able to segment individual cells by using the halos. The halos between cells are detected by thresholding and skeletonization, and gaps in the edges are filled in using Dijkstra’s algorithm. The result of the algorithm was compared with the results of manual segmentation performed by a human expert and evaluated on level of precision of individual cells’ edges. The source code and test data, including the manual segmentation are made available to the public.
Trvalý link: http://hdl.handle.net/11104/0239120