目的 以腹部常规剂量40%ASIR 重建来评价两种基于模型迭代重建算 法(MBIRSTND与MBIRNR40)在低辐射剂量扫 描时的图像质量及辐射剂量减低率。 方法 连续纳入接受腹部增强复查的患 者20例，首次检查行常规剂量(NI=10) 扫描并ASIR40%重建；复查时，采用 低剂量扫描(NI=20)，并采用MBIRSTND 与MBIRNR40重建。客观测量图像噪声值 (SD)和CT值，计算SNR、CNR，比较各 参数之间差异；由２名医师对通图像 进行主观评分，包括图像锐利度、噪 声、伪影和诊断接受度，并进行对比分 析。记录每例患者每次检查的剂量长 度乘积(DLP)和CT剂量指数(CTDIVOL)， 计算剂量减低率。结果 对于腹部脏 器，低剂量MBIRSTND (NI=20)图像与 常规剂量40%ASIR(NI=10)在图像噪声 (SD)、信噪比(SNR)和对比噪声比(CNR) 方面无统计学差异(P＞0.05), 低剂量 MBIRSTND、MBIRNR40重建比40%ASIR图像有 更低伪影和噪声，MBIRNR40图像SNR、CNR 最高，主观评价结果更佳(P＜0.05)。 低剂量扫描的DLP值和CTDIVOL值分别为 (95.56±47.17)mGy*cm和(3.04±1.48) mGy，而常规剂量ASIR 40%重建的DLP值 和CTDIVOL分别为(376.39±160.40)mGy*cm 和(12.16±5.18)mGy，辐射剂量减低约 75%。结论 相比常规剂量扫描(NI=10) ASIR40%重建，低剂量扫描(NI=20) MBIRSTND、MBIRNR40重建在扫描辐射剂量降 低75%的情况下，能有效降低噪声，提高 图像SNR及CNR。

Objective To evaluate the dose reduction and image quality potential in abdominal low-dose CT reconstructed with model-based iterative reconstruction (MBIR) with two different pre-settings (MBIRSTND and MBIRNR40), in comparison with the state-of-the-art adaptive statistical interactive reconstruction (ASIR) in routine-dose CT. Methods Twenty patients underwent two contrast-enhanced abdominal CT scans within two weeks on a 64-slice CT (Discovery CT750HD) with standard-dose (noise index=10HU) and low-dose (noise index=20HU) in the follow-up scan. The routinedose CT was reconstructed with ASIR40 (40% ASIR and 60% FBP mix) and the lowdose CT was reconstructed with both the ASIR40 and the two types of MBIR algorithms. Two radiologists assessed the images blindly according to the sharpness, image noise, diagnostic acceptability and artifacts with 5-point scoring. Image noise, signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) of abdominal organs relative to erector spinae were assessed. The volume CT dose index (CTDIvol), dose-length product (DLP) and dose reduction rate were also determined. Results CTDIvol and DLP for the lowdose CT were (95.56±47.17m)Gy*cm and (3.04±1.48)mGy, respectively, compared with (376.39±160.40)mGy*cm and (12.16±5.18)mGy in the routine-dose CT with 75% reduction (P<0.001). Low-dose (NI=20HU) MBIR images with MBIRSTND setting had similar image noise, SNR and CNR as the routine-dose (NI=10HU) ASIR40 images (all P>0.05). On the other hand, the low-dose MBIR images with MBIRNR40 setting further improved image quality with higher image quality score, reduced image noise and improved SNR and CNR compared with the routine-dose ASIR40 images (all P<0.001). Conclusion MBIR significantly reduced image noise and improved SNR and CNR even at 75% dose reduction, compared with the routine-dose, state-of-the-art ASIR images.