Projekt
Non-femoral head MRI radiomics as a potential imaging biomarker for identifying osteonecrosis of the femoral head in systemic lupus erythematosus patients
Purpose: To explore whether MRI-based radiomics features derived from non-femoral head bone marrow regions differ between symptomatic systemic lupus erythematosus (SLE) patients with and without osteonecrosis of the femoral head (ONFH), and to preliminarily evaluate their potential diagnostic utility. Methods: This re…
Purpose: To explore whether MRI-based radiomics features derived from non-femoral head bone marrow regions differ between symptomatic systemic lupus erythematosus (SLE) patients with and without osteonecrosis of the femoral head (ONFH), and to preliminarily evaluate their potential diagnostic utility. Methods: This retrospective study enrolled SLE patients from the primary institution (temporal split into training/internal validation cohorts) and an external institution using same-vendor scanners (external validation cohort). Radiomics features were extracted from the greater trochanter and ilium on T1-weighted images. A three-step feature selection strategy integrating LASSO, random forest (RF), and support vector machine recursive feature elimination (SVM-RFE) was applied to enhance feature robustness. Clinical, radiomics, and combined models were constructed and evaluated using ROC analysis, calibration curves, and decision curve analysis (DCA). Results: A total of 123 patients were included in the primary cohort and 44 in the external validation cohort. The combined model achieved AUCs of 0.937, 0.889, and 0.861 in the training, internal validation, and external validation cohorts, respectively. The combined model yielded numerically higher AUC than the clinical model across all datasets. The radiomics model alone demonstrated stable discrimination and favorable net benefit in internal and external validation, whereas adding clinical variables did not improve clinical utility. DCA shows that in the validation cohort, the independent radiomics model was more stable in terms of net clinical benefit compared to the combined model. The inter-group differences in the external cohort did not reach a statistically significant level, which might be attributed to insufficient sample size and type II errors. Conclusion: MRI-based radiomics features derived from non-femoral head bone marrow regions differ significantly between symptomatic SLE patients with and without ONFH, suggesting potential imaging biomarker for diagnostic differentiation. These exploratory findings warrant further validation in prospective studies.
Themengebiete
- Gesundheit Gesundheit – Überblick über Forschungsprojekte, Patente und Akteure im TechnologieAtlas.
- Künstliche Intelligenz Künstliche Intelligenz – Überblick über Forschungsprojekte, Förderprojekte und Akteure im TechnologieAtlas.
Hochschulen
- Shanxi Medical University Shanxi Medical University – Hochschule bzw. Forschungseinrichtung mit Aktivitäten in Forschung und Innovation.