Applications of Artificial Intelligence for Pediatric Cancer Imaging.

Area of Science:

• Medical Imaging
• Pediatric Oncology
• Artificial Intelligence

Background:

• Artificial intelligence (AI) is revolutionizing adult medical imaging.
• Its application in pediatric oncology imaging is limited due to rare cancer incidence and insufficient data.
• Rapidly evolving imaging technologies further exacerbate data scarcity for training AI algorithms.

Purpose of the Study:

• To review current AI applications in pediatric cancer imaging.
• To identify challenges and opportunities for AI in this field.
• To explore strategies for enhancing AI clinical translation in pediatric oncology.

Main Methods:

• Review of current AI applications across medical image acquisition, processing, reconstruction, segmentation, diagnosis, staging, and treatment response monitoring.
• Analysis of impediments including anatomical diversity and nonstandardized protocols.
• Identification of opportunities such as accelerated low-dose imaging and automated metric generation.

Main Results:

• AI applications show promise in various aspects of pediatric cancer imaging.
• Significant impediments to clinical translation exist, including data scarcity and protocol variations.
• Leveraging reconstruction algorithms and transfer learning are identified as key opportunities.

Conclusions:

• Despite promising developments, clinical translation of AI in pediatric oncology is limited.
• Addressing data scarcity through multi-institutional sharing and transfer learning is crucial.
• Ethical data privacy practices are essential for advancing AI in rare pediatric cancers.