Optimal sensitivity of fluorescence in situ hybridization (FISH) requi
res bright signals and low background fluorescence. Use of locus-speci
fic probes is especially dependent on high sensitivity. Some tissue pr
eparations show high autofluorescence, masking small or dim signals. W
e have developed a new method for subtracting autofluorescence from di
gital images on a pixel-by-pixel basis. It is based on the observation
that fluorescent labels for FISH have narrower excitation and emissio
n spectra than the chemical components responsible for autofluorescenc
e. Our new approach uses calculation of the ratio of autofluorescence
between multiple color images for correction of autofluorescence in ea
ch individual image. By subtracting autofluorescence components, we we
re able to enhance centromeric signals and make previously indistingui
shable cosmid signals clearly visible. This image-processing approach
to autofluorescence correction may widen the applicability of gene-spe
cific probes in FISH analysis of tumor material. (C) 1995 Wiley-Liss,
Inc.