CE-Net is a context encoder network designed for 2D medical image segmentation. It addresses the issue of spatial information loss in traditional U-Net architectures by incorporating a feature encoder module, a context extractor module, and a feature decoder module. The feature encoder uses a pretrained ResNet block, while the context extractor includes a dense atrous convolution (DAC) block and a residual multi-kernel pooling (RMP) block. These components help capture high-level semantic features and preserve spatial information. The feature decoder restores high-resolution feature maps for accurate segmentation. CE-Net outperforms existing methods in tasks such as optic disc segmentation, retinal vessel detection, lung segmentation, cell contour segmentation, and retinal OCT layer segmentation. It uses a Dice coefficient loss function for better performance in medical image segmentation. The method was validated on multiple datasets, showing improved results compared to U-Net and other state-of-the-art approaches. The proposed CE-Net is effective for both two-class and multi-class segmentation tasks, demonstrating its versatility in medical image analysis.CE-Net is a context encoder network designed for 2D medical image segmentation. It addresses the issue of spatial information loss in traditional U-Net architectures by incorporating a feature encoder module, a context extractor module, and a feature decoder module. The feature encoder uses a pretrained ResNet block, while the context extractor includes a dense atrous convolution (DAC) block and a residual multi-kernel pooling (RMP) block. These components help capture high-level semantic features and preserve spatial information. The feature decoder restores high-resolution feature maps for accurate segmentation. CE-Net outperforms existing methods in tasks such as optic disc segmentation, retinal vessel detection, lung segmentation, cell contour segmentation, and retinal OCT layer segmentation. It uses a Dice coefficient loss function for better performance in medical image segmentation. The method was validated on multiple datasets, showing improved results compared to U-Net and other state-of-the-art approaches. The proposed CE-Net is effective for both two-class and multi-class segmentation tasks, demonstrating its versatility in medical image analysis.