Araki Tetsuro, Sholl Lynette M, Gerbaudo Victor H, Hatabu Hiroto, Nishino Mizuki
Department of Radiology, Brigham and Women's Hospital, 75 Francis St, Boston, 02215 MA.
Department of Pathology, Brigham and Women's Hospital, Boston, MA.
Acad Radiol. 2014 Jun;21(6):733-42. doi: 10.1016/j.acra.2014.02.006.
To determine the intraobserver and interobserver variabilities of thymic measurements on computed tomography (CT) in patients with pathologic diagnosis of thymic hyperplasia or normal thymus.
Thirty-three patients with pathologic diagnosis of thymic hyperplasia (n = 25) or normal thymus (n = 8) who had identifiable thymus gland on CT were retrospectively studied. Two radiologists independently measured thymic size and CT attenuation. Concordance correlation coefficients (CCCs) and Bland-Altman plots were used to assess intraobserver and interobserver agreements.
The intraobserver and interobserver agreements of thymic diameters and the lobe length were moderate, with CCCs ranging from 0.73 to 0.89 and from 0.72 to 0.81, respectively. Higher agreement was noted among patients whose measurements were performed on the same CT image in two independent measurements, with intraobserver CCC ≥ 0.95 for diameters and length. After providing readers with an instruction for consistent selection of CT image for measurements, the intraobserver and interobserver agreements improved, resulting in CCCs ranging from 0.81 to 0.92 and from 0.77 to 0.85 for diameters and length, respectively. Thymic lobe thickness had the least agreement. CT attenuation measurements were highly reproducible, with CCCs ranging from 0.88 to 0.97. In patients with thymic CT attenuation >30 HU (Hounsfield unit), the attenuation measurements were more reproducible with narrower 95% limits of agreement.
Thymic size measurements had moderate-to-high intraobserver and interobserver agreements, when the instruction for consistent selection of images was provided to the readers. CT attenuation was highly reproducible, with higher reproducibility for thymic glands with >30 HU. Awareness of thymic measurement variability is necessary when interpreting measured values of normal thymus and thymic pathology on CT.
确定经病理诊断为胸腺增生或胸腺正常的患者,其胸腺在计算机断层扫描(CT)上测量的观察者内和观察者间变异性。
回顾性研究33例经病理诊断为胸腺增生(n = 25)或胸腺正常(n = 8)且CT上可识别胸腺的患者。两名放射科医生独立测量胸腺大小和CT衰减值。一致性相关系数(CCC)和Bland-Altman图用于评估观察者内和观察者间的一致性。
胸腺直径和叶长度的观察者内和观察者间一致性为中等,CCC分别为0.73至0.89和0.72至0.81。在两次独立测量中对同一CT图像进行测量的患者中,一致性更高,直径和长度的观察者内CCC≥0.95。在为读者提供关于一致选择测量CT图像的指导后,观察者内和观察者间的一致性得到改善,直径和长度的CCC分别为0.81至0.92和0.77至0.85。胸腺叶厚度的一致性最低。CT衰减测量具有高度可重复性,CCC为0.88至0.97。在胸腺CT衰减>30 HU(亨氏单位)的患者中,衰减测量的可重复性更高,一致性界限的95%范围更窄。
当向读者提供一致选择图像的指导时,胸腺大小测量具有中等至高的观察者内和观察者间一致性。CT衰减具有高度可重复性,对于衰减>30 HU的胸腺,可重复性更高。在解释CT上正常胸腺和胸腺病变的测量值时,必须意识到胸腺测量的变异性。