Bowel Perforation Detection

Stylised image of a bowel
Stylised image of a bowel

Academic Lead: Dr Hammad Khan

Clinical Lead: Dr Ravi Bhat

Clinical Area: Perinatal imaging

Partner: Carnax

Early detection of bowel perforation in pre-term babies significantly improves patient outcome. The pathology, which has a missing rate of 25% from literature, has the potential to degenerate within hours, causing morbidity and mortality. 

Using AI technology, the project aims to support clinicians in detecting bowel perforations in pre-term neonates. The AI will review abdominal X-rays to detect intestinal perforation, that will help determine the ongoing care and decision for the patient. The development of the AI algorithm has begun but further images of positive cases of bowel perforation are required to improve the sensitivity and specificity of the AI algorithm.  

The AI tool will be trained on data in the form of anonymised DICOM images abdominal and chest X-Rays on pre-term neonates. Additional data will also be used to address the risk of bias in the development of the algorithm.