A Computed Tomography-based Radiomics Nomogram for Predicting Osteoporotic Vertebral Fractures: A Longitudinal Study

Area of Science:

• Medical Imaging
• Radiology
• Geriatrics

Background:

• Osteoporotic fractures pose a significant health risk to older adults.
• Longitudinal studies on CT-based radiomics for predicting osteoporotic vertebral fractures (OVFs) are limited.

Purpose of the Study:

• To evaluate a CT radiomics-based model for predicting OVFs in a longitudinal cohort.
• To assess the performance of a predictive nomogram incorporating radiomics and muscle features.

Main Methods:

• A longitudinal study of 7906 individuals over 50 without OVFs, followed from 2016-2021.
• Extraction of radiomics features from baseline CT scans, including vertebral CT values, trabecular bone properties, and erector spinae muscle characteristics.
• Cox regression analysis and nomogram development for OVF prediction, validated using ROC curves, calibration curves, and decision curves.

Main Results:

• Vertebral CT value, radiomics score, and erector spinae muscle area were independent predictors of OVF.
• The radiomics nomogram demonstrated good discrimination (C-index = 0.82) and outperformed traditional osteoporosis assessments for 3- and 4-year fracture risk.
• The radiomics nomogram showed clinical utility in decision curve analysis.

Conclusions:

• CT-based bone radiomics is significantly associated with OVF.
• A nomogram integrating radiomics signatures and muscle features provides an effective tool for predicting osteoporotic vertebral fractures.