TDC-Unet: Triple Unet with Dilated Convolution for Medical Image Segmentation

Medical image segmentation is one of the research directions that are interested in recent years. The Unet model is one of the most architecture commonly used for medical image segmentation. However, Unet and Unetbased models still have a drawback that is concentrating only on the last feature output of the convolution unit and forgetting the feature of the previous convolution in the node. In this paper, we propose a new model based on Unet model, called by TDC-Unet that would exploit the intrafeature of the nodes in the Unet architecture. We also apply the Dilated Convolution (DC) and dense connection in the nodes structure. We used four datasets, that cover different modalities of medical image: colonoscopy, dermoscopy, and Magnetic Resonance Imaging (MRI) to evaluate the proposed model. The applications in our experiment are: nuclei segmentation, polyp segmentation, left atrium segmentation, and skin lesion segmentation. The experimental results show that our model achieves better results than the current models.