Viewing and analyzing slide scanner data using CellProfiler (work in progress)
2013 (English)In: European BioImage Analysis Symposium 2013 / [ed] Julien Colombelli, 2013, p. 58-Conference paper, Poster (with or without abstract) (Other academic)
Abstract [en]
Viewing and analyzing slide scanner data using CellProfiler (work in progress)
Petter Ranefall1, Alexandra Pacureanu1, and Carolina Wählby1,2
1Centre for Image Analysis, Department of Information Technology, Uppsala University, and Science for Life Laboratory, Sweden
2Imaging Platform, Broad Institute of Harvard and MIT, Cambridge, MA, USA
The amount of image data in a slide scanner image is usually very large, inducing challenges for image analysis and visualization. We would like to use the strengths and flexibility of CellProfiler for the image analysis, but performing high-resolution image analysis on large slide scanner images is unfortunately not possible due to memory limitations. At the same time, we would like to visualize the results of the analysis in the context of the full tissue slide so that for example phenotypic variations at the sub-cellular level can be related to lower-resolution structures such as vessels, ducts, or tumors in the tissue.
To approach this problem we split the image into smaller tiles that are more suitable for CellProfiler to handle. By keeping track of the coordinates of the tiles we can display the results on the original, full-size image. Our aim is to enable visual examination at multiple resolutions and with the option to toggle results such as segmentation masks and classification results on or off using a “Google Maps” type of view.
In particular, we work with tissue profiling by in situ sequencing of RNA molecules. The tissue samples have to be removed from the microscope for each new sequencing cycle. In this case, and in other applications dealing with repeated staining, there are often differences in the alignment between the imaging rounds. We assume that these misalignments are rigid (only translation and rotation), and we have added an image registration step in order to align the different channels before partitioning the images into tiles suitable for analysis in CellProfiler.
Place, publisher, year, edition, pages
2013. p. 58-
National Category
Medical Image Processing
Research subject
Computerized Image Analysis
Identifiers
URN: urn:nbn:se:uu:diva-209613OAI: oai:DiVA.org:uu-209613DiVA, id: diva2:658701
Conference
European BioImage Analysis Symposium 2013, Barcelona
2013-10-222013-10-222022-01-28Bibliographically approved