ESR Essentials: radiomics-practice recommendations by the European Society of Medical Imaging Informatics

Area of Science:

• Medical Imaging Analysis
• Quantitative Imaging Biomarkers
• Radiology Informatics

Background:

• Radiomics offers valuable insights from medical images, but its complexity hinders clinical adoption.
• Methodological challenges in radiomics analysis impact reproducibility and generalizability.
• Understanding radiomics is crucial for improving clinical decision-making.

Purpose of the Study:

• To provide recommendations and practical examples for good research practices in radiomics.
• To address challenges in radiomics analysis, including feature extraction and model development.
• To enhance the clinical translation of radiomics models.

Main Methods:

• Detailed review of critical steps in radiomics analysis and model development.
• Identification of common errors in data partitioning, model assessment, and calibration.
• Emphasis on model explainability and interpretation of radiomic features.

Main Results:

• Methodological complexities significantly affect radiomic feature values and model performance.
• Errors in the pipeline can reduce reproducibility and impede clinical translation.
• Adherence to best practices is essential for accurate feature extraction and robust model development.

Conclusions:

• Understanding radiomics' methodological complexities is key to rigorous model development.
• Standardized tools and adherence to radiomics-specific checklists ensure methodological rigor.
• Implementing best practices increases the likelihood of successful clinical translation of radiomics models.