Cure8 news brief
Cure8 news brief
Fusing blood tests, CT enterography radiomics, and deep-learning analysis of deep small-bowel endoscopy images may improve detection of active small-bowel Crohn’s disease — a common clinical challenge because many affected segments lie beyond standard colonoscopy reach.
Clinicians who manage Crohn’s disease (especially small-bowel disease), gastroenterology researchers working on imaging and AI, and patients interested in advances in disease activity monitoring.
Researchers used retrospective clinical data and three imaging-related streams: selected laboratory and clinical variables (including CRP and fecal calprotectin), quantitative radiomics from CT enterography, and features plus a binary output from a ResNet-18 model trained on double-balloon enteroscopy images.
They compared single-modality models to combined models and found the fully integrated models performed best by AUC and multiple comparative tests (DeLong, NRI, IDI), and they packaged results as nomograms and used SHAP to explore feature contributions.
How this might be used The approach is primarily a diagnostic/monitoring tool idea: combining systemic biomarkers, cross-sectional imaging, and mucosal images to improve detection of small-bowel activity. The study is retrospective and from a single center; the models would need external validation and prospective testing before clinical use.
This is a retrospective, single-center study using double-balloon enteroscopy images and CT enterography radiomics. Findings are promising but require external validation and prospective testing before changing clinical practice. The article summarizes model-building and internal performance metrics rather than reporting clinical outcomes from prospective use.
Review the original publication for the complete reporting, methods, and context.
This Cure8 brief is based on source text from the linked article. Cure8 is informational only and is not a substitute for professional medical advice, diagnosis, or treatment.