Training radiomics-based CNNs for clinical outcome prediction: Challenges, strategies and findings

Medicine & Life Sciences

• Tomography 55%
• Area Under Curve 53%
• Tumor Burden 46%
• Datasets 25%
• Machine Learning 18%
• Correlation of Data 18%
• Survival 18%
• Head and Neck Neoplasms 16%
• Sample Size 14%
• Risk Assessment 13%
• Learning 11%
• Radiotherapy 11%
• Neoplasms 10%
• Neoplasm Metastasis 10%
• Therapeutics 5%

Engineering & Materials Science

• Computerized tomography 100%
• Tumors 66%
• Feature extraction 24%
• Radiotherapy 20%
• Risk assessment 15%
• Network architecture 15%
• Convolutional neural networks 14%
• Tissue 13%
• Imaging techniques 12%
• Machine learning 11%
• Experiments 11%