Department of Radiology, Massachusetts General Hospital, Boston, 02114, USA.
AJR Am J Roentgenol. 2010 Jul;195(1):78-88. doi: 10.2214/AJR.09.3420.
The objective of this study was to evaluate the image quality and diagnostic performance of simulated low-dose MDCT examinations in patients with Crohn's disease.
Thirty-five MDCT examinations from 25 patients (14 males and 11 females; age range, 16-86 years) with known or suspected Crohn's disease were retrospectively evaluated. The MDCT images (5 mm thickness) were modified by artificially introducing noise using volume CT noise simulation software to simulate low-dose MDCT data acquired with noise indexes (NIs) of 18-35. The 175 MDCT image data sets generated were then randomized and reviewed by two readers to assess the subjective image quality and diagnostic accuracy for confidently detecting Crohn's disease findings on a 5-point scale (1, definitely absent; 5, definitely present). The image quality, diagnostic performance, and radiation dose from the original MDCT examinations served as a reference standard for comparison.
The simulated low-dose CT images with elevated NIs of 18-25 yielded optimal image quality and concordant diagnostic accuracy with standard-dose scans with a mean baseline volume CT dose index of 16 +/- 3.34 (SD) mGy (image quality score, 4-4.9 vs 4.95, respectively; p = 0.4). Images simulating NIs of 30-35 were considered degraded because of excessive image noise and presented lower diagnostic performance (range of image quality score, 3-3.6; weighted kappa = 0.25-0.37; p < 0.001). Diagnostic performance and reader confidence for the determination of Crohn's disease manifestations in cases in which the reference standard was positive were higher at NIs of < or = 25 compared with NI of 30 (reader confidence score, 4.5 vs 3.6, respectively; p < 0.05) and an NI of 35 (3.2, p < 0.05). Compared with the standard-dose examination, a 31-64% reduction in radiation dose was estimated for NI levels of 18-25, which corresponds to image noise of 19-27 HU in subcutaneous fat.
Processed MDCT images with the introduction of noise to simulate low-dose MDCT examinations with NI levels of 18-25 allows substantial dose reduction for CT examinations in Crohn's disease without compromising diagnostic information.
本研究旨在评估模拟低剂量 MDCT 检查在克罗恩病患者中的图像质量和诊断性能。
回顾性分析 25 例（男 14 例，女 11 例；年龄 16-86 岁）已知或疑似克罗恩病患者的 35 例 MDCT 检查。使用体部 CT 噪声模拟软件对 MDCT 图像（5mm 层厚）进行人工引入噪声，以模拟噪声指数（NI）为 18-35 的低剂量 MDCT 数据。然后将生成的 175 组 MDCT 图像数据随机分配给两位读者进行评估，以 5 分制（1，肯定不存在；5，肯定存在）评估对克罗恩病发现的主观图像质量和诊断准确性。原始 MDCT 检查的图像质量、诊断性能和辐射剂量作为比较的参考标准。
NI 为 18-25 的模拟低剂量 CT 图像具有最佳的图像质量和与标准剂量扫描一致的诊断准确性，标准剂量扫描的平均基线容积 CT 剂量指数为 16±3.34（SD）mGy（图像质量评分分别为 4-4.9 和 4.95；p=0.4）。NI 为 30-35 的图像被认为是退化的，因为图像噪声过大，呈现出较低的诊断性能（图像质量评分范围为 3-3.6；加权 kappa=0.25-0.37；p<0.001）。在参考标准为阳性的情况下，NI<或=25 时，用于确定克罗恩病表现的诊断性能和读者信心高于 NI 为 30（读者信心评分分别为 4.5 和 3.6；p<0.05）和 NI 为 35（3.2，p<0.05）。与标准剂量检查相比，NI 为 18-25 时，辐射剂量降低 31-64%，相当于皮下脂肪的噪声为 19-27HU。
在引入噪声以模拟 NI 为 18-25 的低剂量 MDCT 检查的过程中，对克罗恩病的 CT 检查进行大幅度的剂量减少，而不会影响诊断信息。
