Multicolor Fluorescence Imaging with Spectral Unmixing
A comparison of spectral unmixing verses band pass filter based separation
Depicted are bovine pulmonary artery endothelial cells (BPAEC) where Texas Red-X is labeling the F-actin, BODIPY FL (green) is labeling microtubules, and DAPI (blue) is labeling the nucleus.
Filter Based Separation
Below, three “virtual” band pass images are created by summing photon intensity data from a (sub)set of spectral image “slices” whose wavelengths correspond to those that an emission band pass filter chosen for each fluorophore would permit through. These red, green and blue channels are then combined into a final RGB composite image.
Spectral Unmixing
Below, linear spectral unmixing is applied to each pixel to extract relative intensity component of each fluorophore labeling the sample using published emission spectra subsequently normalized for the sample. Spectral map images for each dye are constructed and then combined into a final composite image.
Conclusions
An examination of these two image series illustrates that spectral unmixing methods applied to hyperspectral fluorescence images allow for clearer delineation of the “true” fluorescent component/signal coming from various cellular structures than does band pass filter based approaches, especially when emission spectral overlaps occur among the set of fluorophores used.