Multi-Scale Network Based on Dilated Convolution for Bladder Tumor Segmentation of Two-Dimensional MRI Images

Because of the bladder tumors are often not completely removed during surgery, the recurrence rate after resection is high. In order to separate the bladder tumor from the bladder wall, this paper proposes a multiscale network based on dilated convolution (DMC-Unet). In the DMC-Unet model, we use U-net as the basic network framework. Firstly, the network uses dilated convolutions with different dilation rates to obtain different receptive fields and then realizes the structure of branch downsampling to obtain features of different scales. Secondly, the residual structure is used to extract feature as the sub-module of DMC-Unet, which effectively alleviates the disappearance of the gradient. In addition, a data enhancement strategy is proposed in this paper. We use Gaussian noise, spatial geometric transformation and gamma transformation to amplify the data to avoid overfitting due to the small dataset. Finally, the effectiveness of our method is verified by experiments, high accuracy has been obtained for bladder wall and bladder tumor segmentation aiming at two-dimensional MRI images.