Dual fluorescence labelling is an advanced method to separate two individual specimens in a biological system using confocal microscopy. An inherent problem of this method is fluorescence channel cross-talk, which causes problems for the exact spatial determination and separation of the specimens. Using a parallel fluorescence detection and an image processing technique, based on an image subtraction method, we have developed a very straight forward method for correcting the dual channel fluorescence images. We successfully applied this method to a 3-dimensional cancer spheroid invasion assay and controlled the cross-talk compensation efficiency by a quality parameter.