Artificial intelligence and radiomics in pediatric molecular imaging

Area of Science:

• Medical Imaging Analysis
• Precision Medicine
• Artificial Intelligence in Healthcare

Background:

• Radiomics, the extraction of quantitative features from medical images, has emerged as a powerful tool in precision medicine.
• It aims to discover novel imaging biomarkers for disease diagnosis, prognosis, and molecular subtyping.
• Traditional radiomics relies on hand-engineered features, while neural networks offer an alternative approach.

Purpose of the Study:

• To explore the evolving role and broad opportunities of radiomics and neural networks in pediatric nuclear medicine, radiology, and molecular imaging.
• To highlight the potential of these technologies in enhancing diagnostic and prognostic capabilities.
• To identify key application areas for AI in pediatric medical imaging.

Main Methods:

• Quantitative analysis of medical images using radiomics to extract shape and texture parameters.
• Application of neural networks for direct learning and identification of predictive features from medical images, bypassing hand-engineered features.
• Review of potential applications in pediatric imaging.

Main Results:

• Radiomics has demonstrated success in extracting diagnostic, prognostic, and molecular information from medical images.
• Neural networks show improved performance and reliability by learning predictive features directly, reducing reliance on predefined parameters.
• Significant opportunities exist for AI in automating and augmenting clinical tasks and developing new biomarkers.

Conclusions:

• Radiomics and neural networks represent a significant advancement in quantitative medical image analysis and prognostic modeling.
• These AI-driven approaches hold immense potential for transforming pediatric nuclear medicine and radiology.
• Key applications include intelligent order sets, automated protocoling, improved image acquisition, computer-aided detection, and advanced biomarker development.