Medical image super-resolution (SR) has emerged as an effective means to enhance the resolution of medical images. Nevertheless, many existing methods still face the issue of insufficient representation of high-frequency features. To address this problem, we propose a Transformer-based hierarchical Encoder–Decoder Network (THEDNet). The THEDNet incorporates an advanced transformer to extract features at various hierarchical dimensions. Specifically, the Encoder and Decoder units are equipped with an Efficient Multi-scale Attention (EMA) module for capturing long-range interdependencies among features. By leveraging an enhanced transformer architecture, THEDNet can capture long-range feature interdependencies and create the final high-resolution images. Experiments are conducted on two medical CT image datasets, and comparative results verify the effectiveness of the proposed THEDNet.