
Resolution improvement by 3‐D reconstructions from tilted views in axial tomography and confocal theta microscopy
Author(s) -
Sätzler Kurt,
Eils Roland
Publication year - 1997
Publication title -
bioimaging
Language(s) - English
Resource type - Journals
eISSN - 1361-6374
pISSN - 0966-9051
DOI - 10.1002/1361-6374(199712)5:4<171::aid-bio1>3.0.co;2-k
Subject(s) - optics , optical axis , resolution (logic) , confocal , microscopy , anisotropy , image resolution , physics , microscope , isotropy , artificial intelligence , computer science , lens (geology)
The resolution along the optical axis is much less than the lateral resolution in both confocal and conventional fluorescence light microscopy. To correct for the anisotropy in resolution one may generate tilted views from the same object and subsequently merge these tilted views into one image with improved isotropic resolution. It has been suggested to obtain tilted views in confocal theta microscopy with double observation by illuminating the object from one side and simultaneously detecting the emitted light in the direction parallel to the illumination direction and in a direction with an angle theta to the illumination direction. Alternatively, one might use the usual optical set‐up of a light microscope together with a tilting device to rotate the object under the microscope. With this method the object can be viewed from all directions ( axial tomography ). In this paper we investigate two methods for merging tilted views. If the image sets obtained from the different tilt angles are interpreted as statistic samples of one and the same object, a gain in resolution along the optical axis is accompanied by a significant trade‐off in lateral resolution. An optical interpretation of the image formation in each single object point by retaining the frequencies with the highest amplitudes in the different tilted views provides a gain in axial resolution without significant decrease in lateral resolution. We will describe a quantitative method to determine the gain in resolution dependent on the number and geometry of the tilted views. It will be demonstrated that a previously suggested reconstruction method implying a multiplication of tilted views in the spatial domain does not improve resolution. Simulated reconstructions based on measured point spread functions in axial tomography and confocal theta microscopy with double observation will demonstrate the validity of our reconstruction methods.