Virtual non-calcium subtraction values enable accurate osteoporosis diagnosis via quantitative CT

Virtual non-calcium (VNCa) imaging derived from dual-energy computed tomography (DECT) is increasingly used in the diagnosis of spinal disorders.However, the effectiveness of VNCa technology in assessing vertebral bone status is unclear, especially across different levels of calcium suppression. This study aimed to evaluate the diagnostic performance of VNCa imaging for osteoporosis using quantitative CT (QCT) as the reference standard.196 patients were selected for chest CT energy spectrum mode scanning. The images were transferred to the quantitative CT (QCT) workstation, and the vertebral bodies were divided into osteoporotic and non-osteoporotic groups with reference to the QCT diagnostic standards. The diagnostic performance of osteoporosis was evaluated by comparing the vertebral CT values when calcium suppression index (CaSI) was 25, 50, and 75 and the subtraction value between it and the conventional CT plain value. All ROC analyses were performed within a cluster-adjusted framework to account for the within-patient correlation among the three vertebrae contributed per patient. A total of 587 vertebral bodies (129 osteoporotic and 458 non-osteoporotic) from 196 patients were analyzed. After adjusting for within-patient clustering, the AUC values for diagnosing osteoporosis using VNCa CT values at CaSI 25, 50, and 75 were 0.928 (95% CI: 0.901–0.955), 0.841 (95% CI: 0.801–0.880), and 0.680 (95% CI: 0.629–0.730), respectively. The corresponding AUC values for the subtraction values were 0.960 (95% CI: 0.941–0.979), 0.964 (95% CI: 0.948–0.980), and 0.963 (95% CI: 0.948–0.978). The subtraction value significantly outperformed the corresponding VNCa CT value at all three CaSI levels (all P < 0.0001). Conventional CT values alone yielded an AUC of 0.980 (95% CI: 0.970–0.990). Spectral CT calcium suppression imaging can effectively evaluate bone status. Within a cluster-adjusted analytical framework, the subtraction value demonstrated the highest diagnostic performance among all VNCa-based metrics, with an AUC of 0.964 (95% CI: 0.948–0.980) at CaSI 50, significantly outperforming VNCa CT values at all CaSI levels. Conventional CT values also showed excellent diagnostic performance, supporting their potential as a widely available opportunistic screening tool. External validation in multi-center cohorts is warranted to confirm these findings.