På svenska

Presentation Information     2008-11-24 (15:15)   •  The seminar room at Vi2

Speaker Milan Gavrilovic
Title Applications of spectral angle based classification in microscopy image processing
Abstract Last year a novel method for pixel classification by spectral angles was developed at CBA. The method was originally used for automated and robust detection of colocalization in fluorescence microscopy which was the subject of my previous seminar. The same method was recently modified and applied for solving two common problems in microscopy - cross-talk, or bleed-through, and autofluorescence, also known as background or tissue fluorescence. In particular, an automated method for autofluorescence suppression based on fuzzy classification of pixels will be presented and compared to background subtraction, a classical approach used when images are recorded in two distinct spectral channels.