Diekhoff T, Hermann K G, Pumberger M, Hamm B, Putzier M, Fuchs M
Department of Radiology, Charité - Universitätsmedizin Berlin Campus Mitte, Humboldt-Universität zu Berlin, Freie Universität Berlin, Berlin, Germany.
Department of Radiology, Charité - Universitätsmedizin Berlin Campus Mitte, Humboldt-Universität zu Berlin, Freie Universität Berlin, Berlin, Germany.
Eur J Radiol. 2017 Feb;87:59-65. doi: 10.1016/j.ejrad.2016.12.008. Epub 2016 Dec 12.
Dual-energy computed tomography (DECT) is a recent development for detecting bone marrow edema (BME) in patients with vertebral compression fractures. The aim of this pilot study was to determine the reliability of single-source DECT in detecting vertebral BME using magnetic resonance imaging (MRI) as standard of reference.
Nine patients with radiographic thoracic or lumbar vertebral compression fractures underwent both, DECT on a 320-row single-source scanner and 1.5T MRI. Virtual non-calcium (VNC) images were reconstructed from the DECT volume datasets. Three blinded readers independently scored images for the presence of BME. Only vertebrae with loss of height in radiography (target vertebrae) were included in the analysis. A vertebra was counted as positive if two readers agreed on the presence of BME. Cohen's kappa was calculated for interrater comparison. Intervertebral ratios of target and the reference vertebra were compared for CT attenuation and MR signal intensity in a reference vertebra using Spearman correlation. Signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were calculated.
Fourteen target vertebrae with a radiographic height loss were identified; eight of them showed BME on MRI, while DECT identified BME in 7 instances. There were no false positive virtual non-calcium images, resulting in a sensitivity of 0.88 (0.75-1.0 among all readers) and specificity of 1.0 (0.81-1.0). Interrater agreement was inferior for DECT (κ=0.63-0.89) compared to MRI (κ=0.9-1.0). Intervertebral ratio in VNC images strongly correlated with short-tau inversion recovery (r=0.87) and inversely with T1 (-0.89). SNR (0.2+/- 0.2 in VNC and 16.7+/- 7.3 in STIR) and CNR (0.2+/- 0.3 and 7.1+/- 6.3) values were inferior in VNC.
Detecting BME with single-source DECT is feasible and allows detection of vertebral compression fractures with reasonably high sensitivity and specificity. However, image quality of VNC reconstructions has to be improved to achieve better interrater agreement. Nonetheless, DECT might accelerate the diagnostic work-flow in patients with vertebral compression fractures in the future and reduce the number of additional MRI examinations.
双能计算机断层扫描（DECT）是检测椎体压缩骨折患者骨髓水肿（BME）的一项最新技术。本初步研究的目的是使用磁共振成像（MRI）作为参考标准，确定单源DECT检测椎体BME的可靠性。
9例经X线检查确诊为胸椎或腰椎椎体压缩骨折的患者，分别接受了320排单源扫描仪的DECT检查和1.5T MRI检查。从DECT容积数据集中重建虚拟非钙（VNC）图像。三位不知情的阅片者独立对图像上是否存在BME进行评分。分析仅纳入X线片上有高度丢失的椎体（目标椎体）。如果两位阅片者都认为存在BME，则该椎体计为阳性。计算Cohen's kappa值用于评估阅片者间的一致性。使用Spearman相关性分析比较目标椎体与参考椎体之间的CT衰减和MR信号强度的椎间比值。计算信噪比（SNR）和对比噪声比（CNR）。
共识别出14个X线片上有高度丢失的目标椎体；其中8个在MRI上显示有BME，而DECT仅在7例中识别出BME。VNC图像没有假阳性，灵敏度为0.88（所有阅片者中为0.75 - 1.0），特异性为1.0（0.81 - 1.0）。与MRI（κ = 0.9 - 1.0）相比，DECT的阅片者间一致性较差（κ = 0.63 - 0.89）。VNC图像中的椎间比值与短Tau反转恢复序列（r = 0.87）呈强相关，与T1加权像呈负相关（r = -0.89）。VNC图像的SNR（0.2 ± 0.2）和CNR（0.2 ± 0.3）值低于短Tau反转恢复序列（分别为16.7 ± 7.3和7.1 ± 6.3）。
使用单源DECT检测BME是可行的，能够以较高的灵敏度和特异性检测椎体压缩骨折。然而，必须改善VNC重建图像的质量，以获得更好的阅片者间一致性。尽管如此，DECT未来可能会加快椎体压缩骨折患者的诊断流程，并减少额外的MRI检查次数。
