Bladder cancer is a prevalent and potentially life-threatening disease that requires accurate diagnosis and treatment planning. Medical image segmentation plays a crucial role in the assessment of tumor location, size, and progression. In this paper, We investigated the application of a U-Net based deep learning mode for bladder cancer segmentation. To perform bladder cancer segmentation using the U-Net technique, a diverse dataset of bladder cancer images is collected, comprising various stages and types of bladder cancer. The images are preprocessed to enhance contrast and remove noise, ensuring optimal input quality for the U-Net model. Subsequently, the U-Net model is trained using a large set of annotated images, where pixel-wise annotations serve as ground truth for the segmentation task. Initial tests using five sets of TCIA dataset show that the suggested algorithm achieves an average DSC of 87.28% and an average time of 6.0 minutes without parallelized computation, clearly surpassing other current techniques for bladder segmentation.