Convolutional neural networks (CNNs) have been significantly advanced in the field of medical imaging in recent years. The implementation of U-shaped architectures has proven particularly effective, demonstrating exceptional results across various image segmentation challenges. Despite these achievements, convolutional operations face challenges in capturing broad-scale global context and long-distance semantic relationships. Therefore, the ECLNet is presented in this paper, a converter with an architecture similar to U-Net that was specifically designed for medical image segmentation. With the help of the Transformer architecture, complete global context is extracted by this model through the processing of feature maps produced by the CNN as input sequences. To achieve accurate feature localization, the CNN feature map is combined with the global context information. It is claimed that the Transformer design can be utilized as a reliable encoder for medical image segmentation applications. Its combination with convolutional networks enhances segmentation efficiency by facilitating the restoration of local spatial information. Finally, extensive experiments on one public and one private datasets show that our proposed algorithm performs better than several state-of-the-art methods.