Pediatric Buried Bumper Syndrome: Diagnostic Validity of Transabdominal Ultrasound and Artificial Intelligence.

Abstract

Buried bumper syndrome (BBS) is a severe complication of percutaneous endoscopic gastrostomy (PEG) resulting from overgrowth of gastric mucosa and penetration of the inner holding plate into the gastric wall. The aim of this study was to evaluate the diagnostic value of transabdominal ultrasound (US) in comparison to an artificial intelligence (AI) model for the diagnosis of BBS in children. In this monocentric retrospective study, pediatric US data concerning BBS from a ten-year period (2009-2019) were analyzed. US findings were compared to a clinical multiparameter-based AI model and reference standard endoscopy. Clinical risk factors for the occurrence of pediatric BBS were determined. In n = 121 independent examinations of n = 82patients, the placement of the inner holding plate of the PEG was assessed by US. In n = 18 cases BBS was confirmed. Recall and precision rates were 100 % for US and 88 % for the AI-based assessment. Risk factors for the occurrence of BBS were mobilization problems of the PEG (rs = 0.66, p < 0.001), secretion/exudation (rs = 0.29, p = 0.002), time between 1stPEG placement and US (rs = 0.38, p < 0.001), and elevated leukocyte count (rs = 0.24, p = 0.016). Transabdominal US enables correct, rapid, and noninvasive diagnosis of BBS in pediatric patients. Preceding AI models could aid during diagnostic workup. To avoid unnecessary invasive procedures, US could be considered as a primary diagnostic procedure in suspected BBS. .