Development of a generalized model for kidney depth estimation in the Chinese population: A multi-center study

• Published in: European journal of radiology Vol. 124; p. 108840
• Main Authors: Li, Qian, Pan, Zhongyun, Li, Qiang, Baikpour, Masoud, Cheah, Eugene, Chen, Kai, Li, Wenliang, Song, Yiqing, Zhang, Jingjing, Yu, Lijuan, Zuo, Changjing, Liu, Jianjun, Yang, Aimin, Ding, Zhiling, Li, Juan, Luo, Yongjun, Li, Tiannv, Feng, Yanlin, Yu, Shupeng, Xie, Laiping, Luo, Ganhua, Wang, Qian, Wei, Longxiao, Chen, Yue, Sun, Hua, Lin, Chenghe, Xu, Wengui, Zhao, Wenrui, Peng, Xiang, Wang, Cheng, Han, Xingmin, Ba, Ya, Zhang, Yanjun, Li, Wei, Zhang, Wei, Yang, Hui
• Format: Journal Article
• Language: English
• Published: Ireland Elsevier B.V 01.03.2020
• Subjects: Age; Computed tomography; Height; Kidney depth; Weight; Kidney depth; Height; Computed tomography; Weight; Age
• ISSN: 0720-048X; 1872-7727
• DOI: 10.1016/j.ejrad.2020.108840

•New kidney depth equations were constructed using multiple regression analysis.•New equations had high accuracy for kidney depth prediction, confirmed by CT.•New equations had lower estimation errors compared to other established models.•New equations had generalizability to people from all Chinese regions. To establish an accurate and reliable equation for kidney depth estimation in adult patients from different Chinese geographical regions. This multicenter study enrolled Eastern Asian Chinese patients with abdominal PET/CT scans at 26 imaging centers from six macro-regions across China in 3 years. Age, gender, height, weight, primary disease and its extent on PET scans of the participants were collected as potential predictive factors. Kidney depth on CT, defined as the average of the vertical distances from the posterior skin to the farthest anterior and closest posterior surfaces of each kidney, was measured as the standard reference. The new kidney depth model was constructed using a multiple regression model, and its performance was compared to those of three established models by computing the absolute value of estimation errors in comparison with CT-measured kidney depth. A total of 2502 patients were enrolled and classified into training (n=1653) and testing (n = 849) subsets. In the training subset, two kidney depth models were constructed: Left (cm): 0.013×age+0.117×gender-0.044×height+0.087×weight+7.951, Right (cm): 0.005×age+0.013×gender-0.035×height+0.082×weight+7.266 (weight: kg, height: cm, gender = 0 if female, 1 if male). In the testing subset, one-way analysis of variance showed that the estimation errors of the new models did not significantly differ among the 6 regions. Bland-Altman analysis determined that new equations had lower estimated biases (left: 0.039 cm, right: 0.018 cm) compared with other existing models. The new equations were highly accurate for kidney depth estimation in adults from all over China, with lower estimation errors compared to other established models.