Image fusion is the process by which we combine data from two or more source images to a single image with all the necessary information of the source images. Image fusion methods have found applications in many different areas and fields such as remote sensing, CCD cameras, medical imaging and microscopy. In our case we would like to use image fusion methods to improve images acquired from Light Sheet Fluorescence Microscopy (LSFM) or Selective Plane Illumination Microscopy (SPIM). SPIM is a fluorescence microscopy technique in which the specimen is illuminated by a micrometer thin focused laser sheet, along an optical path orthogonal to the detection axis. Since only the in-focus part of the specimen is exposed to laser light, it allows for optical sectioning of the specimen at high penetration depth and substantially reduces photo-bleaching. However, when dealing with a large or highly scattering sample the quality of the images deteriorates, resulting in loss of information. To increase the information content of the SPIM images and compensate for the penetration depth limitations, due to the light absorption and scattering, we acquire scans from different views (projections) of the sample and then combine them in a way that provides a complete view of the sample. To do that we need to accurately register and fuse these images. Image registration is the process by which different sets of data are transformed into one coordinate system; we define a transform T that will map one image onto another image of a similar or same object in a way that maximizes their alignment. We have previously presented a technique that uses images acquired as mentioned above, and by rigid and non-rigid image registration aligns the SPIM images. In this thesis we present and compare different techniques that fuse the already aligned images to create one final high resolution three dimensional image.
(EN)
University of Crete
