文献类型：期刊文献

英文题名：Performance evaluation of a deep learning-based cascaded HRNet model for automatic measurement of X-ray imaging parameters of lumbar sagittal curvature

作者：Wu, Yuhua[1];Chen, Xiaofei[2];Dong, Fuwen[2];He, Linyang[3];Cheng, Guohua[3];Zheng, Yuwen[1];Ma, Chunyu[1];Yao, Hongyan[4];Zhou, Sheng[4]

第一作者：吴玉泓

通信作者：Zhou, S[1]

机构：[1]Gansu Univ Chinese Med, Clin Med Coll 1, Lanzhou 730000, Gansu, Peoples R China;[2]Gansu Univ Tradit Chinese Med, Gansu Prov Hosp Tradit Chinese Med, Affiliated Hosp 1, Dept Radiol, Lanzhou 730050, Gansu, Peoples R China;[3]Hangzhou Jianpei Technol Co Ltd, Hangzhou 311200, Zhejiang, Peoples R China;[4]Gansu Prov Hosp, Dept Radiol, 204 Donggang West Rd, Lanzhou 730000, Gansu, Peoples R China

第一机构：甘肃中医药大学

通信机构：[1]corresponding author), Gansu Prov Hosp, Dept Radiol, 204 Donggang West Rd, Lanzhou 730000, Gansu, Peoples R China.

年份：2023

外文期刊名：EUROPEAN SPINE JOURNAL

收录：;Scopus(收录号：2-s2.0-85173090755);WOS:【SCI-EXPANDED(收录号:WOS:001074609400001)】；

基金：Hangzhou Jianpei Technology Co., Ltd., employees are gratefully acknowledged for providing practical and technical resources.

语种：英文

外文关键词：Deep learning; Automatic measurement; Lumbar sagittal curvature; Lumbar lordosis angle (LLA); Sacral slope (SS); Radiography

摘要：PurposeTo develop a deep learning-based cascaded HRNet model, in order to automatically measure X-ray imaging parameters of lumbar sagittal curvature and to evaluate its prediction performance.MethodsA total of 3730 lumbar lateral digital radiography (DR) images were collected from picture archiving and communication system (PACS). Among them, 3150 images were randomly selected as the training dataset and validation dataset, and 580 images as the test dataset. The landmarks of the lumbar curve index (LCI), lumbar lordosis angle (LLA), sacral slope (SS), lumbar lordosis index (LLI), and the posterior edge tangent angle of the vertebral body (PTA) were identified and marked. The measured results of landmarks on the test dataset were compared with the mean values of manual measurement as the reference standard. Percentage of correct key-points (PCK), intra-class correlation coefficient (ICC), Pearson correlation coefficient (r), mean absolute error (MAE), mean square error (MSE), root-mean-square error (RMSE), and Bland-Altman plot were used to evaluate the performance of the cascade HRNet model.ResultsThe PCK of the cascaded HRNet model was 97.9-100% in the 3 mm distance threshold. The mean differences between the reference standard and the predicted values for LCI, LLA, SS, LLI, and PTA were 0.43 mm, 0.99 degrees, 1.11 degrees, 0.01 mm, and 0.23 degrees, respectively. There were strong correlation and consistency of the five parameters between the cascaded HRNet model and manual measurements (ICC = 0.989-0.999, R = 0.991-0.999, MAE = 0.63-1.65, MSE = 0.61-4.06, RMSE = 0.78-2.01).ConclusionThe cascaded HRNet model based on deep learning algorithm could accurately identify the sagittal curvature-related landmarks on lateral lumbar DR images and automatically measure the relevant parameters, which is of great significance in clinical application.