Automated image analysis in fluorescence microscopy: From isolated cells to tissues and microarray images
|Year of publication
|Article in Proceedings
|Biophysics of the Genome
|MU Faculty or unit
|Use of computers, robotics and its application
|Laboratory of Optical Microscopy has been working on the automation of image analysis in fluorescence microscopy since mid-1990s. At that time mostly 2D image analysis of FISH stained cell nuclei was explored. The images contained isolated cells (such as blood cells) that did not form many clusters and, therefore, image segmentation (the most difficult part of image analysis) was quite easy to automate. Later on 3D images of FISH (or immunofluorescence) stained cell nuclei acquired in confocal mode were dealt with. Also in this case the cell nuclei were well separated from each other, however somewhat blurred in axial direction due to the inferior resolution along z-axis. Also for this case automated methods have been developed including algorithms for the computation of a mathematical model of cell (or cell nucleus) boundary in 3D. At the end of 1990s the necessity of tissue image analysis arose. In this area, fully automated image segmentation is not possible, hence the effort has been concentrated on semi-automatic approaches with minimal user interaction. Recently, also the demand of processing in vivo image series (of separated cells) has appeared, so automated object tracking algorithms have been investigated. Finally, our laboratory has started to produce also images of microarray slides, which are a special type of fluorescence microscopy images. For this type of images, completely new image processing methods are being developed that try to find grids as well as spots in an automated way. The presentation will be an overview of image analysis methods used in our laboratory for processing of the above-mentioned types of image data.