European journal of radiology · 2026
Clinical-Radiomics hybrid prediction of the Risks of pedicle screw loosening after posterior lumbar fusion.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Han Weitao, Gao Songbo, Han Shun, Zhong Xiaolin, Zhang Penghui, Wu Yuliang, Tian Haotian, Wei Fuxin, Wen Shifeng, Zhao Shen, Ye Wei
Аннотация
OBJECTIVE: We aimed to construct a risk prediction model for PSL after posterior lumbar fusion using machine learning and radiomic methods.
SUMMARY OF BACKGROUND INFORMATION: Posterior lumbar fusion surgery is a routine procedure for the treatment of lumbar degenerative disease (LDD). Pedicle screw loosening (PSL) is a common complication after posterior lumbar fusion that can lead to postoperative pain and infection in the implant area.
METHODS: A total of 304 patients who underwent posterior lumbar fusion for LDD at Sun Yat-sen Memorial Hospital were reviewed in this study. 80 patients at the Seventh Affiliated Hospital of SYSU and Guangzhou First People's Hospital made up the external validation datasets. Preoperative demographic and surgical information was collected. One-year-postoperative radiological follow-up was conducted, and patients were divided into PSL and non-PSL groups. Clinical and surgical information was subjected to Student's t test for feature selection. Lumbar CT images were analyzed using radiomic methods, and PSL prediction models were constructed by machine learning methods. The best model was selected and externally validated on the data from the two other hospitals.
RESULTS: All 304 patients were included. Age (p＜0.001), Preoperative lumbar lordosis (LL) (p = 0.006), mean CT Hounsfield units (p = 0.007), and number of fixed segments (p < 0.001) differed between the two groups. Logistic regression revealed that the number of fixed segments was an independent risk factor (OR = 2.147, p < 0.001). Radiomic features were selected after feature extraction and selection. After model training and testing, the clinical + radiomic model showed acceptable predictive performance (AUC 0.894). Its AUCs in the two external validation datasets were 0.821 and 0.892 respectively.
CONCLUSION: Combining clinical and radiomic features can better predict the risk of PSL after posterior lumbar fusion surgery. This investigation revalidated the risk factors for PSL after posterior lumbar fusion as well.
SUMMARY OF BACKGROUND INFORMATION: Posterior lumbar fusion surgery is a routine procedure for the treatment of lumbar degenerative disease (LDD). Pedicle screw loosening (PSL) is a common complication after posterior lumbar fusion that can lead to postoperative pain and infection in the implant area.
METHODS: A total of 304 patients who underwent posterior lumbar fusion for LDD at Sun Yat-sen Memorial Hospital were reviewed in this study. 80 patients at the Seventh Affiliated Hospital of SYSU and Guangzhou First People's Hospital made up the external validation datasets. Preoperative demographic and surgical information was collected. One-year-postoperative radiological follow-up was conducted, and patients were divided into PSL and non-PSL groups. Clinical and surgical information was subjected to Student's t test for feature selection. Lumbar CT images were analyzed using radiomic methods, and PSL prediction models were constructed by machine learning methods. The best model was selected and externally validated on the data from the two other hospitals.
RESULTS: All 304 patients were included. Age (p＜0.001), Preoperative lumbar lordosis (LL) (p = 0.006), mean CT Hounsfield units (p = 0.007), and number of fixed segments (p < 0.001) differed between the two groups. Logistic regression revealed that the number of fixed segments was an independent risk factor (OR = 2.147, p < 0.001). Radiomic features were selected after feature extraction and selection. After model training and testing, the clinical + radiomic model showed acceptable predictive performance (AUC 0.894). Its AUCs in the two external validation datasets were 0.821 and 0.892 respectively.
CONCLUSION: Combining clinical and radiomic features can better predict the risk of PSL after posterior lumbar fusion surgery. This investigation revalidated the risk factors for PSL after posterior lumbar fusion as well.