NON-RIGID TEMPORAL REGISTRATION OF 2D AND 3D MULTI-CHANNEL MICROSCOPY IMAGE SEQUENCES OF HUMAN CELLS

The analysis of fluorescently tagged proteins in live cells from multi-channel microscopy image sequences requires a registration to a reference frame to decouple the movement and deformation of cells from the movement of proteins. We have developed an intensity-based approach to register 2D and 3D multi-channel microscopy image sequences. This approach directly exploits the image intensities. We have compared the results of our approach with results based on segmented images. Also, we have performed a comparison between a direct registration scheme to the reference frame and an incremental scheme taking into account results from preceding time steps. We have validated our approach based on 3D synthetic images of a simulated cell with known deformation which has been calculated based on an analytic solution of the Navier equation given certain boundary conditions. We have also successfully applied our approach to 2D and 3D real microscopy image sequences