Multi-scale U-net with Edge Guidance for Multimodal Retinal Image Deformable Registration

Keyphrases

• Edge Similarity 100%
• Multimodal Retinal Image 100%
• Edge Guidance 100%
• Deformable Image Registration 100%
• U-Net 100%
• Similarity Loss 66%
• Optical Coherence Tomography Images 33%
• Eye Diseases 33%
• Regularizer 33%
• Loss Function 33%
• Existing Algorithms 33%
• Multi-scale Input 33%
• Competitive Results 33%
• Color Fundus Image 33%
• Squeeze-and-excitation 33%
• Dilated Convolution 33%
• Abnormal Image 33%
• Edge Diffusion 33%
• Spatial Gradient 33%
• Registration Problem 33%
• Scale Layer 33%
• Local Details 33%
• Receptive Field 33%
• Similarity Structure 33%
• Input Layer 33%
• Optical Distortion 33%
• Edge Difference 33%
• Local Cross-correlation 33%
• Matrix Deformation 33%
• Convolution Structure 33%
• Block Transformer 33%
• Quantitative Experiments 33%
• Rigid Registration 33%
• Spatial Transformer Layer 33%
• Unsupervised Learning Techniques 33%
• Optical Centers 33%

Engineering

• Multiscale 100%
• Similarities 100%
• Great Importance 33%
• Ground Truth 33%
• Loss Function 33%
• Receptive Field 33%
• Scale Layer 33%
• Input Layer 33%

Computer Science

• U-Net 100%
• Deformable Registration 100%
• Receptive Field 50%
• Unsupervised Learning 50%