Objective:

To establish radiomic and deep learning models based on contrast-enhanced CT, clinical data, and biochemical indicators for the differential diagnosis of peripheral neuroblastoma (NB) and ganglioneuroblastoma (GNB) in children, highlighting the potential impact on clinical decision-making.

Key Findings:

• The study included 225 patients, with 161 diagnosed with NB and 64 with GNB, achieving a diagnostic accuracy of X% (insert specific metric).
• Conventional imaging methods have limitations in differentiating between NB and GNB, with a sensitivity of Y% and specificity of Z% (insert specific metrics).
• Deep learning and radiomics show promise in improving diagnostic accuracy, with potential increases in sensitivity and specificity.

Interpretation:

The integration of advanced imaging techniques and deep learning may enhance the diagnostic process for pediatric tumors, potentially leading to better treatment outcomes and more personalized care strategies.

Limitations:

• The study was retrospective and may have selection bias, which could limit the applicability of the findings to broader populations.
• Limited generalizability due to the specific patient population and imaging protocols used, suggesting the need for multicenter studies.

Conclusion:

Advanced CT radiomics and deep learning models can provide valuable support in the differential diagnosis of pediatric peripheral neuroblastoma and ganglioneuroblastoma, aiding in personalized treatment strategies.