Development of Image Processing Tools for Correlative Microscopy

Abstract

Correlative microscopy is a useful microscopy technique, where a particular region in a sample is imaged using multiple different microscopy systems. The effectiveness of correlative microscopy is thereby based on combining the advantages of different systems, since each microscopy technique has its strengths and weaknesses. For instance, light microscopy can be used to image large areas but with a compromised resolution, while electron microscopy techniques provide a better resolution but have a smaller field of view. Correlative light and electron microscopy is a common example of correlative microscopy, but the same principle can be applied to various forms of microscopy and analysis techniques.

However, presenting the combined data from different systems is a major challenge in correlative microscopy since the images need to be aligned correctly for analysis. While tools for microscopy data alignment have been developed, a need for a custom software exists. The main benefit of a custom tool is the flexibility it offers since it enables adding and modifying features according to the needs of the users.

In this thesis, image processing tools for a microscopy data alignment and visualization software are developed using MATLAB. The focus on this work is on the image processing side, while the complete software includes a graphical user interface. The algorithms required for the software can be grouped into geometric transformations, image enhancement and edge detection. The theoretical background of these transformations is also presented in the thesis. Furthermore, a tool for combining multiple images into one by a user defined expression is developed. This tool has its own interface, which is developed as a pop-up window.

The resulting software is useful for microscopy image alignment and visualization. The manual alignment is convenient and can be done with a sub-pixel accuracy. Most of the image enhancement algorithms have been developed for visualization purposes, which helps in the analysis of the data. The developed image merging tool provides creative possibilities for processing the images and creating new information from existing data. The major issue on the image processing side of the software is the slowness of the edge detection algorithm. Otherwise, the software shows good results in its current state and is a potential platform for further development.