Magnetic resonance (MR) and computer tomography (CT) images are two typical types of medical images that provide mutually-complementary information for accurate clinical diagnosis and treatment. However, obtaining both images may be limited due to some considerations such as cost, radiation dose and modality missing. Recently, medical image synthesis has aroused gaining research interest to cope with this limitation. In this paper, we propose a bidirectional learning model, denoted as dual contrast cycleGAN (DC-cycleGAN), to synthesis medical images from unpaired data. Specifically, a dual contrast loss is introduced into the discriminators to indirectly build constraints between MR and CT images by taking the advantage of samples from the source domain as negative sample and enforce the synthetic images fall far away from the source domain. In addition, cross entropy and structural similarity index (SSIM) are integrated into the cycleGAN in order to consider both luminance and structure of samples when synthesizing images. The experimental results indicates that DC-cycleGAN is able to produce promising results as compared with other cycleGAN-based medical image synthesis methods such as cycleGAN, RegGAN, DualGAN and NiceGAN. The code will be available at https://github.com/JiayuanWang-JW/DC-cycleGAN.