Medical image segmentation of abdominal organs in CT scans remains challenging due to complex anatomical structures, low contrast, and variable organ boundaries. To address these limitations, we propose GDCA-TransUNet, built upon the TransUNet architecture, which integrates Dual Cross Attention (DCA) and Attention Gates (AGs). The DCA module enhances feature fusion in skip connections by modeling channel-wise and spatial dependencies between encoder and decoder features, while AGs suppress irrelevant background regions, ensuring a sharper focus on relevant organ boundaries. Evaluated on the Synapse multi-organ CT dataset, GDCA-TransUNet achieves a mean Dice score of 79.98%, outperforming TransUNet (77.48%) and SwinUNet (79.13%). Notably, it achieves 61.97% Dice for the pancreas, demonstrating significant improvements for small and complex organs. By enhancing multi-organ segmentation, particularly for intricate structures, our GDCA-TransUNet has the potential to improve the precision of radiotherapy planning and surgical interventions, where accurate organ delineation is crucial for successful outcomes and personalized treatment strategies.