Interpretable Deep Learning Model for Pediatric Strangulated Small Bowel Obstruction on CT: A Multicenter Study

In brief

This multicenter study presents an interpretable deep learning model combining CT imaging and clinical data to distinguish strangulated from simple small bowel obstruction in children. The multi-instance learning approach aims to improve diagnostic accuracy for this time-sensitive surgical emergency.

• Deep learning can integrate CT imaging with clinical data to distinguish strangulated from simple small bowel obstruction in children.
• Multi-instance learning models show promise for improving diagnostic accuracy in pediatric SBO, potentially reducing unnecessary surgeries.
• Combining imaging and clinical features enhances prediction of strangulation, a time-sensitive surgical emergency in pediatric patients.
• Multicenter validation demonstrates generalizability of AI models for pediatric abdominal emergencies across different institutions.

Written by the GCMD Library team from the article.