Magnetic resonance imaging (MRI) and computed tomography (CT) are essential diagnostic tools that offer complementary insights: MRI is superior in soft tissue contrast, while CT provides high-resolution structural details. However, most multimodal imaging studies have focused on the brain, leaving a significant gap in the analysis of abdominal organs. To address this issue, we propose a novel deep learning framework for abdominal MR-CT image fusion, utilizing generative adversarial network (GAN) and frequency transform that is well suited to capture curved structures in abdominal anatomy. The proposed model focuses on the image fusion stage enhanced by a GAN strategy. Extensive experiments have been conducted with a family of frequency transforms such as wavelets and contourlets. During the training phase, the GAN architecture is strengthened with two discriminators that evaluate the fused image based on gradient and texture fidelity. The effectiveness of the method is rigorously evaluated using several objective no-reference metrics, including entropy, correlation coefficient, spatial frequency, and mean gradient, consistently outperforming baseline fusion methods .