Abstract: To improve the accuracy of Glioma image segmentation and ensure that the edge information of the entire tumor can be obtained more accurately, based on U-shaped convolutional neural network (UNet), a dilated convolutional pyramid model is designed to capture the context of the image at multiple scales to obtain more features. Combined with the efficient channel attention network (ECA) attention module, the model could pay more attention to the most informative channel features, while suppressing those unimportant channel features. The experimental results show that the FLAIR sequence has advantages in segmenting the entire tumor. The intersection over union (IoU) and Dice coefficient (DSC) scores of the designed model in the FLAIR sequence can reach 0.93 and 0.86, respectively, which are 0.07 and 0.05 higher than the original UNet network. It can be concluded that the proposed model effectively obtains more edge information, thereby improving the segmentation accuracy of the entire tumor region in glioma images.