Nested Recurrent Residual Unet (NRRU) on GAN (NRRG) for Cardiac CT Images Segmentation Task
Kening Le, Zeyu Lou, Tian Xiaolin · 2021
With the development of technology, medical imaging technology plays a more and more important role in the processing of diagnosing. As a popular method of digital image processing, convolutional neural network especially U-Net take lots of medical image segmentation tasks. Thus, kinds of models or methods about U-Net are proposed in these years. In this paper, we proposed two new network architectures. One is a new member of U-Net family and created by hybridizing two networks, R2U-Net and Unet++, then the new “U-Net” was named Nested Recurrent Residual U-Net (NRRU). The other network was named Nested Recurrent Residual Unet-GAN (NRRG). Its architecture is based on Pixel-to-Pixel, where NRRU acts as the generative network. After that, I test the performance of NRRG on the dataset from MICCAI 2017 Multi-Modality Whole Heart Segmentation Challenge (MM-WHS 2017), the results show that the new architecture performance good.