Radiomics in PET/CT: Current Status and Future AI-Based Evolutions.

Area of Science:

• Medical Imaging
• Radiomics
• Artificial Intelligence

Background:

• Radiomics is rapidly evolving in Positron Emission Tomography/Computed Tomography (PET/CT) imaging.
• Numerous challenges exist in study design, data acquisition, segmentation, feature calculation, and modeling.
• Some issues are specific to PET/CT and SPECT/CT imaging.

Purpose of the Study:

• To provide an update on the current status of radiomics in PET/CT imaging.
• To discuss future perspectives and potential solutions for identified challenges.
• To focus on pitfalls and recommendations specific to PET/CT imaging.

Main Methods:

• Review of current radiomics methodologies in PET/CT.
• Identification of common and modality-specific pitfalls.
• Exploration of artificial intelligence (AI) and deep learning applications.

Main Results:

• Methodological solutions and recommendations exist to improve study quality and reproducibility.
• AI, including deep learning, shows significant potential for automation in radiomics workflows.
• Challenges remain for the full realization of AI's impact in radiomics.

Conclusions:

• Addressing identified pitfalls is crucial for advancing radiomics in PET/CT.
• AI and deep learning are poised to revolutionize radiomics, offering automation and enhanced analysis.
• Continued rapid advancements are expected in the field of AI-driven radiomics.