Image Synthesis in Nuclear Medicine Imaging with Deep Learning: A Review

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Radiology

Background:

• Nuclear medicine imaging (NMI) is crucial for disease diagnosis and monitoring.
• Current NMI faces challenges in image quality and accessibility during treatment.
• Limitations include low-dose scan artifacts and radiopharmaceutical-specific information loss.

Purpose of the Study:

• To review deep learning methods for synthetic NMI generation.
• To enhance the interpretability and utility of NMI protocols.
• To explore the potential of synthetic images in improving patient outcomes.

Main Methods:

• Analysis of 30 recent publications on deep learning in NMI.
• Discussion of advanced image generation algorithms.
• Evaluation of synthetic image generation techniques for low-dose and enhanced sensing.

Main Results:

• Deep learning models create synthetic NMI closely resembling real images.
• Synthetic images significantly improve diagnostic accuracy with low-dose scans.
• Deep learning facilitates multimodal imaging integration, broadening NMI applications.

Conclusions:

• Deep learning shows significant potential in advancing NMI.
• Synthetic image generation is key to overcoming NMI limitations.
• Improved NMI through synthetic imaging can lead to better patient outcomes.