Radiomics in PET Imaging:: A Practical Guide for Newcomers.

Area of Science:

• Medical Imaging
• Radiology
• Quantitative Imaging

Background:

• Radiomics, particularly in PET imaging, shows promise for predicting lesion characteristics, patient prognosis, and treatment response.
• Handcrafted features in radiomics have demonstrated significant potential in analyzing medical images.

Purpose of the Study:

• To present a comprehensive checklist for designing robust radiomic studies.
• To provide an overview of the radiomic analysis pipeline and data harmonization techniques.
• To compare handcrafted radiomics with deep learning approaches for PET image characterization.

Main Methods:

• Development of a checklist for reliable radiomic study design.
• Description of the radiomic analysis pipeline steps.
• Exploration of data harmonization strategies.
• Comparative analysis of handcrafted features versus deep learning.

Main Results:

• Promising results reported in the literature for handcrafted radiomic features in PET imaging.
• Discussion of advantages and limitations of both handcrafted and deep learning methods.
• Guidance on critical appraisal of radiomic study articles.

Conclusions:

• Radiomics offers valuable tools for PET image analysis, aiding in lesion characterization and patient management.
• Careful study design and data harmonization are crucial for reliable radiomic findings.
• Both handcrafted and deep learning methods have roles in advancing radiomics, with distinct advantages and limitations.