Fluorescence in situ Hybridization technology is a commonly used tool to detect chromosome aberrations, which are often pathologically significant. Since manual FISH analysis is a tedious and time-consuming procedure, reliable and robust automated image acquisition and analysis are in demand. Under high magnification objective lenses such as 60x and 100x, the depth of field will often be too small and the FISH probes may not always lie in the same focal plane. A statistical variance based automated FISH analysis method is developed in order to address this problem. On a stack of slices at consecutive image planes with a step size d, the statistical variance alone the z-axis is calculated to form a 2-D matrix. Since pixels shift dramatically to high intensity at FISH probe location, the probes will manifest high peak values in the matrix. A computer-aided detection scheme based on top-hat transform is applied to the matrix to detect FISH probe signals. This study demonstrates a simple and robust method for FISH probe detection as well as a way of 2- D representation of 3-D data.