Radiomics: A primer for the radiation oncologist

Area of Science:

• Radiomics and medical imaging analysis.
• Quantitative feature extraction from medical images.
• Application of informatics in predictive modeling.

Background:

• Radiomics leverages medical imaging to extract quantitative features for patient phenotyping.
• Various imaging modalities and software packages can be utilized for radiomics analysis.

Purpose of the Study:

• To explain the major steps of a radiomics analysis pipeline.
• To present studies on radiomics in the context of radiation therapy.

Main Methods:

• A literature review was conducted using Medline via PubMed.
• Search terms included "radiotherapy", "radiation oncology", and "radiomics".
• Hand searching of reference lists was performed in July 2019.

Main Results:

• A standard radiomics workflow involves five steps: imaging/segmentation, data curation, feature extraction, selection, and modeling.
• Published radiomics studies in radiation oncology cover lung, head and neck, esophageal, rectal, pancreatic cancers, and brain metastases.
• The quality of retrospective radiomics studies is varied, and clinical translation is pending.

Conclusions:

• Radiomics holds significant potential for predicting clinical outcomes and personalizing radiation therapy.
• The field is nascent and evolving, necessitating improvements in bias reduction and multicenter studies.
• Further research is needed to develop robust and generalizable radiomics models for clinical use.