One of the most important problems in biomedical image analysis is the low amount of data and the cost of accessing to the marked data by researchers. In order to provide a solution to this problem, microscopic fluorescence in situ hybridization (FISH) images are synthesized with generative adversarial network in this paper. The generative adversarial network is trained to synthesize FISH images from mask images. The trained model was implemented on 150 test images and the performance of the model both was presented with visual results and evaluated quantitatively by calculating the performance metrics. By evaluating the synthesized FISH images in terms of image quality and structural features, it is observed that they can be used to provide a solution to the problem of the lack of data.