On Sensitivity and Robustness of Normalization Schemes to Input Distribution Shifts in Automatic MR Image Diagnosis

Computer Science

• Input Distribution 100%
• Deep Learning Model 100%
• Normalization Scheme 100%
• Reconstructed Image 66%
• Image Analysis 33%
• Generalization Performance 33%
• Performance Degradation 33%
• Art Performance 33%
• Image Reconstruction 33%
• Data Distribution 33%
• Performance Model 33%
• Medical Application 33%
• Gold Standard 33%
• Medical Imaging 33%
• Reconstruction Process 33%
• Batch Normalization 33%
• Image Artifact 33%

Keyphrases

• Automatic MR 100%
• Group Normalization 50%
• Soft-tissue Contrast 25%