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

当前位置： X-MOL 学术 › Pediatr. Nephrol. › 论文详情

Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)

Deep learning imaging features derived from kidney ultrasounds predict chronic kidney disease progression in children with posterior urethral valves
Pediatric Nephrology ( IF 2.6 ) Pub Date : 2022-07-22 , DOI: 10.1007/s00467-022-05677-0
John K Weaver _{1,

2} , Karen Milford ₃ , Mandy Rickard ₃ , Joey Logan _{1,

4,

5} , Lauren Erdman _{6,

7} , Bernarda Viteri ₈ , Neeta D'Souza ₁ , Andy Cucchiara ₄ , Marta Skreta _{6,

7} , Daniel Keefe ₃ , Salima Shah _{1,

4} , Antoine Selman _{1,

4} , Katherine Fischer ₁ , Dana A Weiss ₁ , Christopher J Long ₁ , Armando Lorenzo ₃ , Yong Fan ₉ , Greg E Tasian _{1,

4,

10,

11}

Affiliation

Division of Pediatric Urology, Children's Hospital of Philadelphia, Philadelphia, PA, USA.
Department of Urology, Rainbow Babies and Children's Hospital, Case Western Reserve University School of Medicine, Cleveland, OH, USA.
Division of Urology, Hospital for Sick Children, Toronto, ON, Canada.
Center for Pediatric Clinical Effectiveness, The Children's Hospital of Philadelphia, Philadelphia, PA, USA.
Translational Research Informatics Group, The Children's Hospital of Philadelphia, Philadelphia, PA, USA.
Center for Computational Medicine, Hospital for Sick Children Research Institute, Toronto, ON, Canada.
Department of Computer Science, University of Toronto, Toronto, ON, Canada.
Division of Nephrology, Children's Hospital of Philadelphia, Philadelphia, PA, USA.
Department of Radiology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA.
Departments of Surgery and Biostatistics, Epidemiology, Perelman School of Medicine, University of Pennsylvania, & Informatics, Philadelphia, PA, USA.
Surgery and Epidemiology, , The Children's Hospital of Philadelphia, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA, 19104, USA.

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.

中文翻译：

来自肾脏超声的深度学习成像特征可预测患有后尿道瓣膜的儿童的慢性肾脏病进展

背景

我们试图利用深度学习从产后肾脏超声中提取解剖特征，并评估其在预测患有后尿道瓣膜 (PUV) 的男孩慢性肾脏病 (CKD) 进展的风险和时间方面的表现。我们假设这些特征比单独的肌酐最低点等临床特征更好地预测 CKD 进展。

方法

我们对 1990 年至 2021 年在两个儿科卫生系统接受治疗的 PUV 男孩进行了一项回顾性队列研究。使用深度学习模型从最初的出生后肾脏超声图像中提取肾脏特征。使用随机生存森林构建了三个事件时间预测模型。成像模型包括深度学习成像特征，临床模型包括临床数据，集成模型结合了成像特征和临床数据。建立了单独的模型，以包含 6 个月、1 年、3 年和 5 年可用的时间依赖性临床数据。

结果

分析中包括了 225 名患者。所有模型均表现良好，C 指数为 0.7 或更高。临床模型在所有时间点都优于成像模型，最低肌酐推动了临床模型的性能。将 6 个月成像模型（C 指数 0.7；95% 置信区间 [CI] 0.6，0.79）与 6 个月临床模型（C 指数 0.79；95% CI 0.71，0.86）相结合，得出 6 个月的集成模型的表现优于任一模型（C 指数 0.82；95% CI 0.77，0.88）。