Deep learning imaging features derived from kidney ultrasounds predict chronic kidney disease progression in children with posterior urethral valves

• Published in: Pediatric nephrology (Berlin, West) West), 2023-03, Vol.38 (3), p.839-846
• Main Authors: Weaver, John K., Milford, Karen, Rickard, Mandy, Logan, Joey, Erdman, Lauren, Viteri, Bernarda, D’Souza, Neeta, Cucchiara, Andy, Skreta, Marta, Keefe, Daniel, Shah, Salima, Selman, Antoine, Fischer, Katherine, Weiss, Dana A., Long, Christopher J., Lorenzo, Armando, Fan, Yong, Tasian, Greg E.
• Format: Article
• Language: English
• Subjects: Care and treatment; Child; Chronic kidney failure; Creatinine; Deep Learning; Development and progression; Disease Progression; Humans; Infant; Kidney - diagnostic imaging; Kidney diseases; Machine learning; Male; Medicine; Medicine & Public Health; Methods; Nephrology; Original Article; Pediatrics; Prediction models; Renal Insufficiency, Chronic - diagnostic imaging; Retrospective Studies; Ultrasonic imaging; Ultrasound imaging; Urethra - diagnostic imaging; Urethral Obstruction; Urology; What’s New in Chronic Kidney Disease
• ISSN: 0931-041X; 1432-198X; 1432-198X
• DOI: 10.1007/s00467-022-05677-0

Background We sought to use deep learning to extract anatomic features from postnatal kidney ultrasounds and evaluate their performance in predicting the risk and timing of chronic kidney disease (CKD) progression for boys with posterior urethral valves (PUV). We hypothesized that these features would predict CKD progression better than clinical characteristics such as nadir creatinine alone. Methods We performed a retrospective cohort study of boys with PUV treated at two pediatric health systems from 1990 to 2021. Features of kidneys were extracted from initial postnatal kidney ultrasound images using a deep learning model. Three time-to-event prediction models were built using random survival forests. The Imaging Model included deep learning imaging features, the Clinical Model included clinical data, and the Ensemble Model combined imaging features and clinical data. Separate models were built to include time-dependent clinical data that were available at 6 months, 1 year, 3 years, and 5 years. Results Two-hundred and twenty-five patients were included in the analysis. All models performed well with C-indices of 0.7 or greater. The Clinical Model outperformed the Imaging Model at all time points with nadir creatinine driving the performance of the Clinical Model. Combining the 6-month Imaging Model (C-index 0.7; 95% confidence interval [CI] 0.6, 0.79) with the 6-month Clinical Model (C-index 0.79; 95% CI 0.71, 0.86) resulted in a 6-month Ensemble Model that performed better (C-index 0.82; 95% CI 0.77, 0.88) than either model alone. Conclusions Deep learning imaging features extracted from initial postnatal kidney ultrasounds may improve early prediction of CKD progression among children with PUV. Graphical abstract A higher resolution version of the Graphical abstract is available as Supplementary information