In recent years, hybrid CNN-Transformer architectures have significantly advanced the field of medical image segmentation. However, a common limitation of existing methods is their predominant focus on the spatial domain, which often overlooks the rich information encapsulated within the frequency domain. This oversight proves particularly detrimental when segmenting medical images, which are characterized by inherent difficulties like low-contrast regions and textural homogeneity, where spatial features alone are often insufficient. To address these challenges, we propose Synergistic Spatial-Frequency Learning for Medical Image Segmentation. This architecture consists of: (1) Spatial-wise Encoder that employs Transformer to capture global representations while leveraging dynamic convolution to model fine-grained local features; and (2) Spatial-Frequency Fusion Mechanism that integrates spatial and frequency-domain features, facilitating robust and effective representation learning. We conduct extensive experiments on three widely used medical image segmentation benchmarks (ISIC, Synapse, and ACDC), and the results demonstrate the superior performance and strong generalization ability of the proposed framework.