European spine journal : official publication of the European Spine Society, the European Spinal Deformity Society, and the European Section of the Cervical Spine Research Society · 2024
CT-based radiomics predicts adjacent vertebral fracture after percutaneous vertebral augmentation.
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.
Yang Jin, Zhang Shu-Bao, Yang Shuo, Ge Xiao-Yong, Ren Chang-Xu, Wang Shan-Jin
Аннотация
BACKGROUND: Adjacent vertebral fracture (AVF) is a frequent complication following percutaneous vertebral augmentation (PVA). While radiomics is widely utilized in the field of spinal medicine, its application for assessing the risk of AVF in post-PVA patients remains limited.
OBJECTIVE: We aim to develop and validate predictive models using machine learning algorithms for radiomics and clinical risk factors to assess the risk of AVF after PVA.
MATERIALS AND METHODS: This retrospective study included 158 patients with osteoporotic vertebral compression fractures who underwent PVA at our hospital, of which 48 patients had AVF within 2 years. The patients were divided into train and test cohorts in a ratio of 7:3. Radiomics features of the surgically intervened vertebrae were extracted from CT images, and selected using Mann-Whitney U-test and LASSO regression to construct a radiomic signature. Machine learning algorithms (SVM and LR) were then employed to integrate the radiomics signature with clinical data to develop predictive models. The performance of the model was assessed using Receiver Operating Characteristic (ROC) curves and calibration curves.
RESULTS: Nine optimal radiomics features were selected to form the radiomics model, while five clinical features were identified for the clinical model. The AUCs of the radiomics, clinical, and combined models developed using the SVM algorithm were 0.77, 0.77, and 0.83 on the test cohort, and those of the LR algorithm were 0.78, 0.81, and 0.86.
CONCLUSION: Radiomics and machine learning modeling using postoperative CT images demonstrate noteworthy capability in assessing the risk of AVF following PVA.
OBJECTIVE: We aim to develop and validate predictive models using machine learning algorithms for radiomics and clinical risk factors to assess the risk of AVF after PVA.
MATERIALS AND METHODS: This retrospective study included 158 patients with osteoporotic vertebral compression fractures who underwent PVA at our hospital, of which 48 patients had AVF within 2 years. The patients were divided into train and test cohorts in a ratio of 7:3. Radiomics features of the surgically intervened vertebrae were extracted from CT images, and selected using Mann-Whitney U-test and LASSO regression to construct a radiomic signature. Machine learning algorithms (SVM and LR) were then employed to integrate the radiomics signature with clinical data to develop predictive models. The performance of the model was assessed using Receiver Operating Characteristic (ROC) curves and calibration curves.
RESULTS: Nine optimal radiomics features were selected to form the radiomics model, while five clinical features were identified for the clinical model. The AUCs of the radiomics, clinical, and combined models developed using the SVM algorithm were 0.77, 0.77, and 0.83 on the test cohort, and those of the LR algorithm were 0.78, 0.81, and 0.86.
CONCLUSION: Radiomics and machine learning modeling using postoperative CT images demonstrate noteworthy capability in assessing the risk of AVF following PVA.