Combining phase contrast and immunofluorescence images using geometric hashing

Tim Becker, Sandra Schultz, Daniel H. Rapoport, Amir Madany Mamlouk

Abstract

The analysis of in vitro cultured stem cells is a challenging task. One particularly important question pertains the relationship between proliferation and differentiation. Here, we introduce an image-based method to combine information about cellular genealogies with an analysis of the protein expression of cells in a culture dish. The method uses a time-series of microscopical images and a immunofluorescence (IF) image. The time-series can be used to obtain the genealogical nexus, while the IF-image displays the expression of a chosen marker-protein within each cell. The task was then to find an algorithm, which automatically maps the last image of the time-series onto the immunofluorescence image. Our solution to this problem is to use a cell detection algorithm in the time-lapse images and a cell nuclei detection in the IF-images to determine the position of the cells. Then, the cell positions are used to match the images with a geometric hashing based method. The robustness of the implemented algorithm is demonstrated using microscopical reference data. In addition, the algorithm was used to estimate the displacement of the cell nuclei (IF image) relatively to the cell shape position, i.e. the centroids in the time-lapse data.

Original languageEnglish
Title of host publication2013 IEEE 10th International Symposium on Biomedical Imaging
Number of pages4
PublisherIEEE
Publication date22.08.2013
Pages906-909
Article number6556622
ISBN (Print)978-1-4673-6456-0, 978-1-4673-6454-6
ISBN (Electronic) 978-1-4673-6455-3
DOIs
Publication statusPublished - 22.08.2013
Event2013 IEEE 10th International Symposium on Biomedical Imaging: From Nano to Macro - San Francisco, United States
Duration: 07.04.201311.04.2013
Conference number: 98449

Fingerprint

Dive into the research topics of 'Combining phase contrast and immunofluorescence images using geometric hashing'. Together they form a unique fingerprint.

Cite this