Artificial intelligence for nuclear medicine in oncology

Area of Science:

• Nuclear Medicine
• Artificial Intelligence
• Oncology

Background:

• Artificial intelligence (AI) is increasingly integrated into nuclear medicine for oncology applications.
• Few studies examine nuclear medicine from an AI perspective, despite AI's growing use.
• Nuclear medicine images present unique challenges due to low spatial resolution and high quantitativeness.

Purpose of the Study:

• To provide a perspective on nuclear medicine through the lens of artificial intelligence.
• To categorize AI applications in nuclear medicine.
• To highlight challenges and future directions for AI in nuclear medicine.

Main Methods:

• Review of AI applications in nuclear medicine.
• Categorization of AI by purpose: assisted interpretation (CADe/CADx), additional insight (radiomics/radiogenomics), and augmented imaging.
• Discussion of AI's historical use predating deep learning.

Main Results:

• AI applications in nuclear medicine are categorized into three main areas: assisted interpretation, additional insight, and augmented imaging.
• AI has been utilized in medical imaging prior to the advent of deep learning.
• Specific AI applications include computer-aided detection/diagnosis, gene/prognosis prediction, and image generation.

Conclusions:

• AI offers significant potential to enhance nuclear medicine oncology.
• Key challenges for practical AI implementation include inter-facility harmonization and addressing the 'black box' nature of AI explanations.
• Further research is needed to overcome these hurdles for widespread adoption.