Synthetic data in radiological imaging: current state and future outlook

Area of Science:

• Radiology
• Artificial Intelligence
• Medical Imaging

Background:

• Developing AI for radiology faces significant data limitations, including high costs, privacy concerns, and low disease prevalence.
• Acquiring sufficient and representative annotated patient datasets is a major hurdle for AI deployment in medical imaging.

Purpose of the Study:

• To summarize research trends and practical applications of synthetic data for AI in radiology.
• To explore techniques for generating synthetic data, their applications, and quality control measures.

Main Methods:

• Review of current research trends in synthetic data generation for radiological AI.
• Discussion of various synthetic data generation techniques and their application areas.
• Analysis of quality control and evaluation methods for synthetic imaging data.

Main Results:

• Synthetic data offers advantages over patient data, including reduced harm, lower costs, and improved scalability.
• Various techniques exist for generating synthetic radiological data, with specific applications and quality assessment challenges.
• Current methods for evaluating synthetic imaging data are discussed.

Conclusions:

• Synthetic data holds significant potential to address data availability gaps in radiological AI.
• Further research and development are necessary to fully realize the benefits of synthetic data in medical imaging.