Clinical Application of "Three-Low" Technique Combined with Artificial Intelligence Iterative Reconstruction Algorithm in Aortic CT Angiography
-
摘要:目的 探讨“三低”(低辐射剂量、低对比剂用量及低对比剂流速)技术联合人工智能迭代算法(artificial intelligence iterative reconstruction, AIIR)在主动脉CT血管成像中的应用价值。方法 前瞻性纳入33例主动脉CT血管造影(CT angiography, CTA)的患者,按复查时间先后分为A、B 两组。A组为对照组(100 kV, 0.8 mL/kg, 5 mL/s);B组为“三低”组(70 kV, 0.5 mL/kg, 3 mL/s)。A组使用Karl迭代重建图像,B组分别使用Karl和AIIR重建得到B1和B2组。测量3组升主动脉、降主动脉、腹主动脉、左髂动脉及右髂动脉的CT值和SD值、计算信噪比(signal-to-noise ratio, SNR)和对比噪声比(contrast-to-noise ratio, CNR)。同时对图像质量行主观评分。记录A、B组辐射剂量参数。结果 3组各管腔节段CT值、SD值、SNR及CNR差异均有统计学意义(P<0.001)。B2组CT值、SNR、CNR高于B1组,SD值低于B1组,差异均有统计学意义(P<0.017)。B2组与A组的CT值差异无统计学意义(P>0.017),各管腔节段的SD值、SNR和CNR均优于A组(P<0.017)。3组图像主观评分差异有统计学意义(P<0.05),A组与B2组差异无统计学意义(P>0.017)。B组辐射剂量、对比剂用量较A组分别降低84.14%、37.08%。结论 “三低”联合AIIR算法可以获得和常规剂量扫描相当的主动脉CTA图像质量,而患者的辐射剂量、对比剂用量及对比剂流速都明显降低。
-
关键词:
- 人工智能 /
- 主动脉 /
- 体层摄影术,X线计算机
Abstract:Objective To explore the application value of the "three-low" technique (low radiation dose, low contrast agent dosage and low contrast agent flow rate) combined with artificial intelligence iterative reconstruction (AIIR) in aortic CT angiography (CTA).Methods A total of 33 patients who underwent aortic CTA were prospectively enrolled. Based on the time of their follow-up examinations, the imaging data were divided into Group A and Group B, with Group A being the control group (100 kV, 0.8 mL/kg, 5 mL/s) and Group B being the "three-low" technique group (70 kV, 0.5 mL/kg, 3 mL/s). In group A, the images were reconstructed by Karl iterative algorithm. Group B was divided into B1 and B2 subgroups, with their images being reconstructed by Karl iterative algorithm and AIIR, respectively. The CT and SD values of the ascending aorta, descending aorta, abdominal aorta, left common iliac artery and right common iliac artery were measured, and the signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were calculated. The subjective scoring of image quality was performed. The radiation dose parameters were documented.Results Differences in the CT value, SD value, SNR and CNR of the three groups were statistically significant (P<0.001). The CT value, SNR and CNR of group B2 were significantly higher than those of group B1, while the SD value of group B2 was significantly lower than that of group B1 (P<0.017). There was no significant difference between the CT values of group A and those of group B2 (P>0.017). The SD values, SNR and CNR in group B2 were better than those in group A (P>0.017). There was significant difference in the subjective evaluation of image quality among the three groups (P<0.05), but there was no significant difference between group A and group B2 (P>0.017). The radiation dose and contrast medium dosage in group B decreased 84.14% and 37.08%, respectively, compared with those of group A.Conclusion With the "three-low" technique combined with AIIR algorithm, the image quality of aortic CTA obtained is comparable to that of conventional dose scanning, while the radiation dose, contrast agent dosage and contrast agent flow rate of patients are significantly reduced.-
Keywords:
- Artificial intelligence /
- Aorta /
- Tomography, X-ray computer
-
主动脉疾病是一种发病率及致死率较高的危急重症 [1]。CT血管造影(CT angiography, CTA)已成为主动脉大血管病变诊断、治疗前评估以及术后随访的常用方法。然而,由于主动脉扫描范围宽,对主动脉术后及主动脉瘤随访患者需多次行CTA复查,患者的辐射损伤、对比剂肾病(contrast-media induced nephropathy, CIN)和血管破裂风险会明显增高 [2-4] 。研究表明,使用低管电压技术可降低CTA扫描的辐射剂量,在保证血管图像质量的同时进一步降低对比剂的注射流速和总量[5]。然而,降低管电压会导致图像噪声增加[6]。目前,多种图像重建算法被相继提出,包括基于混合迭代、模型迭代 [7-8]的重建算法等。以上方法均无法很好地减少低剂量下的条纹伪影。目前,一种基于人工智能技术的迭代重建算法(artificial intelligence iterative reconstruction, AIIR)被提出,通过该算法重建出来的图像能够有效抑制噪声[9]。但该算法提升低剂量CTA图像质量的能力仍待评估。因此,本研究将采用AIIR联合使用“三低”技术(低管电压、低对比剂流速与总量)行主动脉CTA扫描,与常规剂量下的主动脉CTA图像质量进行对比,探讨其临床应用价值。
1. 资料与方法
1.1 临床资料
前瞻性纳入2020年1月–2021年3月于四川大学华西医院行主动脉CTA检查的33例患者,男28例,女5例,年龄(59.09±15.14)岁,体质量指数(body mass index, BMI)为(24.31±2.94) kg/m2,其中BMI>24 kg/m2有16例。纳入标准:主动脉夹层术后及主动脉瘤不定期复查的患者。按复查时间先后分为两组,首次复查为A组(对照组,n=33),第二次复查为B组(“三低”组,n=33)。排除标准:①运动伪影重,图像质量无法诊断;②对比剂过敏及严重心肾功能不全;③血液动力学不稳定。本研究通过我院生物医学伦理审查委员会批准(2019年审742号)。
1.2 AIIR重建算法
AIIR结合深度学习(deep learning, DL)技术和模型迭代重建(model-based iterative reconstruction, MBIR)算法的特点,针对MBIR中的正则化项在降噪过程中会改变图像噪声纹理的缺点,AIIR采用一种基于卷积神经网络的DL模型替换MBIR中的正则化项,以实现高效降噪[10]。利用DL模型强大的数据学习能力,由AIIR重建出来的图像不仅能够有效减少条纹伪影,而且还拥有与原始数据相似的噪声及纹理特征,提升图像质量。
AIIR中的DL模型训练样本由大量平扫及多期相增强图像数据组成。所有训练样本图像均由MBIR算法重建获得。为实现对不同剂量条件下的图像去噪,首先在原始数据的投影域上加入不同等级的高斯噪声,以模拟出不同剂量条件下的低剂量图像[11];随后在模型训练过程中,将常规剂量的图像与相对应的模拟低剂量图像同时输入卷积神经网络中,DL能够识别出图像中的信号与噪声,并有效抑制图像中的噪声。
1.3 CT扫描技术及参数
所有患者均采用320排螺旋CT(上海联影 uCT960+)行主动脉CTA扫描。A组管电压100 kV,B组管电压70 kV,两组其余参数相同,管电流自动调节,旋转时间0.5 s,螺距0.9937,重建层厚、层间距均为0.5 mm。A组用Karl迭代重建算法。B组分别使用Karl和AIIR重建得到B1和B2组。使用高压注射器经患者肘前静脉注入非离子型对比剂碘美普尔(含碘400 mg I/mL)及生理盐水。注射方案[12]:A组对比剂总量0.8 mL/kg,注射流率5 mL/s,B组对比剂总量0.5 mL/kg,注射流率3 mL/s;两组对比剂注射完毕后,以相同流速再注入30 mL生理盐水。
1.4 图像后处理及图像分析
1.4.1 图像后处理
将主动脉CTA图像传入后处理工作站(uWS-CT)重建,包括曲面重建(curved plannar reconstruction, CPR)、容积再现(volume rendering, VR)及最大密度投影(maximum intensity projection, MIP)等,同时将B1组图像在探索者平台(ulnnovation-CT)进行AIIR重建得到B2组图像。
1.4.2 客观评价
分别在升主动脉,降主动脉(气管分叉下2 cm),腹主动脉(腹腔干层面),左、右髂动脉(髂动脉分叉下2 cm)以及同层面脊柱旁肌肉勾画80 mm2的感兴趣区(region of interest, ROI),测量各血管管腔的CT值和SD值,ROI选择避开金属支架及运动引起的伪影层面、钙化斑块等。计算信噪比(signal-to-noise ratio, SNR)及对比噪声比(contrast-to-noise ratio, CNR),计算公式:SNR血管=CT血管/SD血管;CNR=(CT血管−CT肌肉)/SD肌肉。
1.4.3 主观评价
由两位高年资放射科诊断医师用双盲法分别对3组CTA图像进行主观评分。按5分法评价,1分:无法满足诊断,图像质量极差,噪声及伪影严重,血管边界不清晰;2分:诊断困难,图像质量欠佳,噪声及伪影重,血管边界模糊;3分:基本满足临床诊断,图像质量中等,噪声及伪影一般;4分:满足临床诊断,图像质量好,噪声小,有少量伪影,血管边界光滑;5分:满足临床诊断,图像质量极好,无伪影,噪声小,血管边界光滑。以图像主观评分≥3分视为符合临床诊断需求[13-14]。
1.5 辐射剂量分析
CT设备自动生成两组患者的辐射指标:剂量长度乘积(dose length product, DLP)、容积CT剂量指数(volume CT dose index, CTDIvol),计算有效辐射剂量(effective dose, ED),ED=DLP×K(主动脉转换系数K=0.015 mSv·mGy−1·cm −1)[15]。
1.6 统计学方法
计量资料用
$ \bar{x} $ ±s表示,分类变量用频数表示。CT值、SD值、SNR、CNR及主观评分比较采用Friedman检验,α=0.05;若差异有统计学意义,再采用Wilcoxion符号秩和检验进行组间两两比较,并对检验水准进行校正(事后两两比较调整检验水准=原检验水准/比较次数,本研究中比较次数为3次,则校正检验水准为 0.05/3≈0.017);2名医师对3组图像主观评分的一致性采用Kappa分析,辐射剂量指标(CTDIvol、DLP、ED)及对比剂用量比较采用独立样本t检验,α=0.05。2. 结果
2.1 图像CT值
结果见表1。3组各血管管腔节段的CT值差异均有统计学意义(P<0.01)。B2组的CT值高于B1组,差异有统计学意义(P<0.017)。
表 1 三组图像CT值比较Table 1. Comparison of the CT values of the three groups of imagesAortic lumen segment Group A (n=33) Group B1 (n=33) Group B2 (n=33) χ2 P Ascending aorta/HU 449.60±61.65 417.23±89.82# 443.36±99.54 14.97 0.001 Descending aorta/HU 441.59±63.37 439.76±73.49# 466.32±85.00 11.09 0.004 Abdominal aorta/HU 433.52±70.33 424.68±77.83# 460.57±89.31 22.06 <0.001 Left common iliac artery/HU 415.80±68.47 409.82±84.29# 435.39±96.76 11.88 0.003 Right common iliac artery/HU 411.06±65.19 396.11±100.63# 438.22±100.40 12.06 0.002 CT values is the attenuation value of vascular CT. Group A: Control group; Group B1: Low-dose Karl iterative reconstruction group; Group B2: Low-dose AIIR group. # P<0.017, vs. Group B2. 2.2 图像SD值
结果见表2。3组各血管管腔节段的SD值差异均有统计学意义(P<0.001)。A组与B2组的SD值均低于B1组,差异有统计学意义(P<0.017);A组的SD值均高于B2组,其中升主动脉、降主动脉及腹主动脉的差异有统计学意义(P<0.017)。
表 2 三组图像SD值比较Table 2. Comparison of the SD values of the three groups of imagesAortic lumen segment Group A (n=33) Group B1 (n=33) Group B2 (n=33) χ2 P Ascending aorta 15.40±3.06*, # 31.11±8.27 13.44±4.73* 48.96 <0.001 Descending aorta 15.51±2.72*, # 38.15±12.62 13.64±3.66* 53.88 <0.001 Abdominal aorta 15.32±2.44*, # 39.01±10.70 13.45±2.51* 52.61 <0.001 Left common iliac artery 16.65±4.26* 36.48±11.92 15.99±7.81* 46.79 <0.001 Right common iliac artery 16.43±4.75* 41.82±13.71 14.84±4.00* 50.73 <0.001 SD values is background noise. Group A, Group B1, and Group B2 denote the same as those in table 1. *P<0.017, vs. Group B1; # P<0.017, vs. Group B2. 2.3 图像SNR比较
结果见表3。3组各血管管腔节段的SNR差异均有统计学意义(P<0.001)。A组与B2组的SNR均高于B1组,差异有统计学意义(P<0.017);A组的SNR均低于B2组,其中降主动脉及腹主动脉的差异有统计学意义(P<0.017)。
表 3 三组图像SNR比较Table 3. Comparison of SNR of the three groups of imagesAortic lumen segment Group A (n=33) Group B1 (n=33) Group B2 (n=33) χ2 P Ascending aorta 30.12±6.37* 14.76±8.50 34.47±14.72* 45.52 <0.001 Descending aorta 29.38±6.94*, # 12.40±3.58 36.07±10.23* 52.91 <0.001 Abdominal aorta 28.94±6.22*, # 11.46±2.80 34.99±7.92* 52.91 <0.001 Left common iliac artery 26.77±8.87* 12.13±3.74 31.18±13.59* 47.09 <0.001 Right common iliac artery 26.86±8.48* 10.46±4.48 31.25±10.13* 51.33 <0.001 SNR: Signal-to-noise ratio. Group A, Group B1, and Group B2 denote the same as those in table 1. *P<0.017, vs. Group B1; # P<0.017, vs. Group B2. 2.4 图像CNR比较
结果见表4。3组各血管管腔节段的CNR差异均有统计学意义(P<0.001)。A组与B2组的CNR均高于B1组,差异有统计学意义(P<0.017);A组的CNR均低于B2组,其中腹主动脉,左、右髂总动脉的差异有统计学意义(P<0.017)。
表 4 三组图像CNR比较Table 4. Comparison of the CNR of the three groups of imagesAortic lumen segment Group A (n=33) Group B1 (n=33) Group B2 (n=33) χ2 P Ascending aorta 28.83±6.75* 12.21±4.37 30.57±9.27* 48.64 <0.001 Descending aorta 28.10±6.32* 13.02±4.25 32.45±8.68* 47.52 <0.001 Abdominal aorta 29.51±7.85*, # 13.97±3.54 36.19±7.94* 49.52 <0.001 Left common iliac artery 27.15±7.59*, # 12.96±4.69 32.19±8.78* 51.46 <0.001 Right common iliac artery 26.79±7.40*, # 12.61±5.29 32.40±8.98* 51.46 <0.001 CNR: Contrast-to-noise ratio. Group A, Group B1, and Group B2 denote the same as those in table 1. *P<0.017, vs. Group B1; #P<0.017, vs. Group B2. 2.5 图像质量主观评价
结果见图1~图3。两位诊断医师对A、B1、B2图像质量评价的Kappa值分别为0.89、0.84及0.80。3组图像质量的主观评分差异有统计学意义(医师1:χ2=34.64,P<0.001,医师2:χ2=40.78, P<0.001),A组和B2组评分均高于B1组(P<0.017)。三组图像质量的主观评分均≥3分。
图 3 患者男,65岁,BMI为25.65 kg/m2,升主动脉瘤,最大径6.4 cm(箭头所指)Figure 3. The patient was a 65-year-old male with a body mass index of 25.65 kg/m2. He had ascending aortic aneurysm with a maximum diameter of 6.4 cm (indicated by the arrow)A-C: Curved plannar reconstruction (CPR), volume rendering (VR) and maximum density projection (MIP) images were reconstructed by the control group's iterative algorithm reconstruction; D-F: CPR, VR and MIP images were reconstructed by low-dose B1 group iterative algorithm; G-I: CPR,VR and MIP images were reconstructed by low-dose B2 group AI optimization technology iterative algorithm.2.6 辐射剂量与对比剂用量
B组的辐射剂量、对比剂用量较A组分别降低约84.14%、37.08%,差异有统计学意义(P<0.05)。见表5。
表 5 两组患者辐射剂量及对比剂用量比较Table 5. Comparison of radiation dose index and contrast agent dosage between two groupsItem Group A (n=33) Group B (n=33) t P CTDIvol/mGy 8.82±0.73 1.39±0.12 57.13 <0.001 DLP/mGy∙cm 664.23±69.70 105.60±11.14 45.47 <0.001 ED/mSv 9.96±1.05 1.58±0.17 45.46 <0.001 Contrast agent dosage 54.64±8.63 34.38±5.43 11.41 0.002 CTDIvol: Volume CT dosimetry index; DLP: Dose length product; ED: Effective dose. Group A: Control group; Group B: Low-dose group. 3. 讨论
随着CT技术临床应用的增加,患者因辐射损伤而诱发癌症的风险逐渐引起放射工作者的关注[16]。有研究表明,每增加1 mSv的X射线有效辐射剂量,恶性肿瘤的发病率将会增加0.05‰[17]。降低管电压能够有效减少辐射剂量,同时提高血管与背景的对比度,因此可进一步减少对比剂的用量[5, 18]。在本研究中,B组辐射剂量与对比剂用量较A组分别减少84.14%、37.08%。 “三低”组主动脉各管腔内的CT值均>395 HU,同时联合AIIR算法使图像噪声明显降低,图像质量均满足临床诊断要求。与5 mL/s流速的常规注射方案相比,本研究采用3 mL/s的注射流速还能减少部分血管弹性差的患者发生血管破裂的风险[5, 19]。
近年来,基于人工智能框架的图像质量优化技术已备受关注[20]。该技术通过训练深度神经网络来学习从一个数据集到另一个数据集的复杂分布,以实现噪声与图像的分离,因此在低剂量CT图像去噪方面有很大的潜力[6, 21]。AIIR结合人工智能技术与模型迭代的优点,在重建过程中,利用模型迭代技术对投影数据进行重建,同时采用人工智能技术识别并抑制图像数据中的噪声信号,以此优化图像质量。AIIR不仅能够有效减少图像的噪声与条纹伪影,同时得益于人工智能技术的应用,使得重建后的图像具有较好的真实感。因此,运用AIIR技术进行图像重建能够在降低患者所受辐射剂量的情况下,保证图像质量。
为减少因降低管电压对图像质量产生的影响,本研究使用AIIR重建算法优化低剂量主动脉CTA的图像质量,并与常规主动脉CTA的图像进行对比。低剂量AIIR重建图像的噪声更低,SNR和CNR更高,表明AIIR算法具有强大的降噪能力。主观评价结果也表明低剂量AIIR重建图像质量与常规剂量Karl重建图像质量相当。值得注意的是,低剂量AIIR重建图像的5分占比高于常规剂量Karl重建图像,表明两位诊断医师更倾向于AIIR算法重建出来的图像。此外,本研究发现,在相同剂量条件下,AIIR重建图像的CT值显著高于Karl重建图像,这可能与AIIR算法在训练过程中所使用的训练样本有关。在AIIR中,训练样本图像是由MBIR算法重建所得。有研究表明[22],MBIR重建图像的CT值高于Karl重建图像,因此,AIIR重建图像的CT值也将高于Karl重建图像。但由于AIIR重建图像拥有更优的SNR和CNR,因此CT值的升高不会影响到病灶的可检测性及诊断准确性。
本研究尚存在一些不足之处:①本研究样本量较少,后续将加大样本量做进一步的研究;②由于本研究采用70 kV低剂量扫描,对主动脉支架置换所得图像质量存在一定的影响,后续考虑使用人工智能算法结合去金属伪影技术进行研究做进一步的探讨。
综上所述,“三低”技术联合AIIR重建算法行主动脉CTA图像扫描,在不限制患者BMI的条件下,不仅可以改善低剂量图像噪声,提高图像质量,还可以大幅降低辐射剂量、对比剂用量及对比剂注射流速。
* * *
利益冲突 所有作者均声明不存在利益冲突
-
图 3 患者男,65岁,BMI为25.65 kg/m2,升主动脉瘤,最大径6.4 cm(箭头所指)
Figure 3. The patient was a 65-year-old male with a body mass index of 25.65 kg/m2. He had ascending aortic aneurysm with a maximum diameter of 6.4 cm (indicated by the arrow)
A-C: Curved plannar reconstruction (CPR), volume rendering (VR) and maximum density projection (MIP) images were reconstructed by the control group's iterative algorithm reconstruction; D-F: CPR, VR and MIP images were reconstructed by low-dose B1 group iterative algorithm; G-I: CPR,VR and MIP images were reconstructed by low-dose B2 group AI optimization technology iterative algorithm.
表 1 三组图像CT值比较
Table 1 Comparison of the CT values of the three groups of images
Aortic lumen segment Group A (n=33) Group B1 (n=33) Group B2 (n=33) χ2 P Ascending aorta/HU 449.60±61.65 417.23±89.82# 443.36±99.54 14.97 0.001 Descending aorta/HU 441.59±63.37 439.76±73.49# 466.32±85.00 11.09 0.004 Abdominal aorta/HU 433.52±70.33 424.68±77.83# 460.57±89.31 22.06 <0.001 Left common iliac artery/HU 415.80±68.47 409.82±84.29# 435.39±96.76 11.88 0.003 Right common iliac artery/HU 411.06±65.19 396.11±100.63# 438.22±100.40 12.06 0.002 CT values is the attenuation value of vascular CT. Group A: Control group; Group B1: Low-dose Karl iterative reconstruction group; Group B2: Low-dose AIIR group. # P<0.017, vs. Group B2. 表 2 三组图像SD值比较
Table 2 Comparison of the SD values of the three groups of images
Aortic lumen segment Group A (n=33) Group B1 (n=33) Group B2 (n=33) χ2 P Ascending aorta 15.40±3.06*, # 31.11±8.27 13.44±4.73* 48.96 <0.001 Descending aorta 15.51±2.72*, # 38.15±12.62 13.64±3.66* 53.88 <0.001 Abdominal aorta 15.32±2.44*, # 39.01±10.70 13.45±2.51* 52.61 <0.001 Left common iliac artery 16.65±4.26* 36.48±11.92 15.99±7.81* 46.79 <0.001 Right common iliac artery 16.43±4.75* 41.82±13.71 14.84±4.00* 50.73 <0.001 SD values is background noise. Group A, Group B1, and Group B2 denote the same as those in table 1. *P<0.017, vs. Group B1; # P<0.017, vs. Group B2. 表 3 三组图像SNR比较
Table 3 Comparison of SNR of the three groups of images
Aortic lumen segment Group A (n=33) Group B1 (n=33) Group B2 (n=33) χ2 P Ascending aorta 30.12±6.37* 14.76±8.50 34.47±14.72* 45.52 <0.001 Descending aorta 29.38±6.94*, # 12.40±3.58 36.07±10.23* 52.91 <0.001 Abdominal aorta 28.94±6.22*, # 11.46±2.80 34.99±7.92* 52.91 <0.001 Left common iliac artery 26.77±8.87* 12.13±3.74 31.18±13.59* 47.09 <0.001 Right common iliac artery 26.86±8.48* 10.46±4.48 31.25±10.13* 51.33 <0.001 SNR: Signal-to-noise ratio. Group A, Group B1, and Group B2 denote the same as those in table 1. *P<0.017, vs. Group B1; # P<0.017, vs. Group B2. 表 4 三组图像CNR比较
Table 4 Comparison of the CNR of the three groups of images
Aortic lumen segment Group A (n=33) Group B1 (n=33) Group B2 (n=33) χ2 P Ascending aorta 28.83±6.75* 12.21±4.37 30.57±9.27* 48.64 <0.001 Descending aorta 28.10±6.32* 13.02±4.25 32.45±8.68* 47.52 <0.001 Abdominal aorta 29.51±7.85*, # 13.97±3.54 36.19±7.94* 49.52 <0.001 Left common iliac artery 27.15±7.59*, # 12.96±4.69 32.19±8.78* 51.46 <0.001 Right common iliac artery 26.79±7.40*, # 12.61±5.29 32.40±8.98* 51.46 <0.001 CNR: Contrast-to-noise ratio. Group A, Group B1, and Group B2 denote the same as those in table 1. *P<0.017, vs. Group B1; #P<0.017, vs. Group B2. 表 5 两组患者辐射剂量及对比剂用量比较
Table 5 Comparison of radiation dose index and contrast agent dosage between two groups
Item Group A (n=33) Group B (n=33) t P CTDIvol/mGy 8.82±0.73 1.39±0.12 57.13 <0.001 DLP/mGy∙cm 664.23±69.70 105.60±11.14 45.47 <0.001 ED/mSv 9.96±1.05 1.58±0.17 45.46 <0.001 Contrast agent dosage 54.64±8.63 34.38±5.43 11.41 0.002 CTDIvol: Volume CT dosimetry index; DLP: Dose length product; ED: Effective dose. Group A: Control group; Group B: Low-dose group. -
[1] QIU P, LI Y, LIU K, et al. Prescreening and treatment of aortic dissection through an analysis of infinite-dimension data. BioData Min,2021,14(1): 24. DOI: 10.1186/s13040-021-00249-8
[2] ROMPEL O, GLOCKLER M, JANKA R, et a1. Third-generation dual-source 70-kVp chest CT angiography with advanced iterative reconstruction in young children: Image quality and radiation dose reduction. Pediatr Radiol,2016,46(4): 462–472. DOI: 10.1007/s00247-015-3510-x
[3] FAUCON A L, BOBRIE G, CLÉMENT O. Nephrotoxicity of iodinated contrast media: From pathophysiology to prevention strategies. Eur Radiol,2019,116: 231–241. DOI: 10.1016/j.ejrad.2019.03.008
[4] YACOUB B, STROUD R E, PICCINI D. et al Measurement accuracy of prototype non-contrast, compressed sensing-based, respiratory motion-resolved whole heart cardiovascular magnetic resonance angiography for the assessment of thoracic aortic dilatation: Comparison with computed tomography angiography. J Cardiovasc Magn Reson,2021,23(1): 7. DOI: 10.1186/s12968-020-00697-x
[5] 李万江, 袁元, 李真林, 等. “三低”剂量与图像重建算法在CT血管成像应用中的研究进展. 中国医疗设备,2020,35(10): 82–86. DOI: 10.3969/j.issn.1674-1633.2020.10.017 [6] 王曼, 王怡宁, 于敏, 等. 人工智能成像优化技术在冠状动脉CT血管成像的初步应用研究. 中华放射学杂志,2020,54(5): 460–466. DOI: 10.3760/cma.j.cn112149-20191219-00990 [7] CHEN H, ZHANG Y, ZHANG W, et al. Low-dose CT via convolutional neural network. Biomed Opt Express,2017,8(2): 679–694. DOI: 10.1364/BOE.8.000679
[8] 贾德召, 王一婧, 马跃, 等. 全模型及混合型迭代重建算法对冠状动脉钙化积分的影响. 中国医学影像学杂志,2020,28(4): 241–245. DOI: 10.3969/j.issn.1005-5185.2020.04.001 [9] 游永春, 李万江, 钟思华, 等. AIIR重建算法对70 kVp低管电压扫描主动脉CTA图像质量影响的研究. 中国医疗设备,2021,36(10): 75–79. DOI: 10.3969/j.issn.1674-1633.2021.10.018 [10] LIU J, ZHANG Y, ZHAO Q, et al. Deep iterative reconstruction estimation (DIRE): Approximate iterative reconstruction estimation for low dose CT imaging. Phys Med Biol,2019,64(13): 135007. DOI: 10.1088/1361-6560/ab18db
[11] ZABIĆ S, WANG Q, MORTON T, et al. A low dose simulation tool for CT systems with energy integrating detectors. Med Phys,2013,40(3): 031102. DOI: 10.1118/1.4789628
[12] CHEN Y, LIU Z, LI M, et al. Reducing both radiation and contrast doses in coronary CT angiography in lean patients on a 16-cm wide-detector CT using 70 kVp and ASiR-V algorithm, in comparison with the conventional 100-kVp protocol. Eur Radiol,2019,29(6): 3036–3043. DOI: 10.1007/s00330-018-5837-9
[13] 王明, 王怡宁, 于敏, 等. AI成像优化联合迭代算法在 “双低"主动脉CTA的初步应用. 放射学实践,2018,33(10): 1009–1016. [14] STEFANIE M, CARLO N, JULIAN L, et al. Effect of automated tube voltage selection, integrated circuit detector and advanced iterative reconstruction on radiation dose and image quality of 3rd generation dual-source aortic CT angiography, an intra-individual comparison. Eur J Radiology,2016,85(5): 972–978. DOI: 10.1016/j.ejrad.2016.02.021
[15] 蒲进, 夏春潮, 赵飞, 等. 双源CT大螺距联合智能调制及迭代重建技术在主动脉夹层成像中的应用. 中华放射学与防护杂志,2019,39(1): 6–10. [16] SHARARA S M, MONNIN S R, RUBIO M, et al. Radiation dose burden of CT angiography be reduced while still accurately diagnosing etiology of acute chest pain? Curr Probl Cardiol,2021,46(4): 100766. DOI: 10.1016/j.cpcardiol.2020.100766
[17] CHEUNG B M Y. Coronary CT angiography and subsequent risk of myocardial infarction. N Engl J Med,2019,380(3): 299–300. DOI: 10.1056/NEJMc1816189
[18] CHEN Y, ZHANG X, XUE H, et al. Head and neck angiography at 70 kVp with a third-generation dual-source CT system in patients: Comparison with 100 kVp. Neuroradiology,2017,59(11): 1071–1081. DOI: 10.1007/s00234-017-1901-4
[19] DING S, MEYSTRE N R, CAMPEANU C, et al. Contrast media extravasations in patients undergoing computerized tomography scanning: A systematic review and meta-analysis of risk factors and interventions. JBI Database System Rev Implement Rep,2018,16(1): 87–116. DOI: 10.11124/JBISRIR-2017-003348
[20] KAHN C E. From images to actions: Opportunities for artificial intelligence in radiology. Radiology,2017,285(3): 719–720. DOI: 10.1148/radiol.2017171734
[21] SINGH G, AL'AREF S J, VAN ASSEN M, et al. Machine learning in cardiac CT: Basic concepts and contemporary data. J Cardiovasc Comput Tomogr,2018,12(3): 192–201. DOI: 10.1016/j.jcct.2018.04.010
[22] HAJDU S D, DANIEL R T, MEULI R A, et al. Impact of model-based iterative reconstruction (MBIR) on image quality in cerebral CT angiography before and after intracranial aneurysm treatment. Eur J Radiol,2018,102: 109–114. DOI: 10.1016/j.ejrad.2018.03.011
-
期刊类型引用(9)
1. 孙其中,孟婷,彭俪颖,韩义成,王锡明. 深度学习全模型迭代算法联合60kVp扫描用于头颈部CT血管成像. 中国医学影像技术. 2025(04): 520-524 . 百度学术
2. 韦有永,王甜甜,罗颖薇,卢林玉,丁艳萍,姚国庆,李清恋,官晓晖. 胸部增强CT联合深度学习全模型迭代算法用于支气管动脉成像. 中国医学影像技术. 2025(04): 530-534 . 百度学术
3. 蔡晓嘉,韩锦涛,刘晶,杨帆,王琦,李如迅. 深度学习全模型迭代算法用于卵巢肿瘤术前低剂量CT. 中国医学影像技术. 2025(04): 539-542 . 百度学术
4. 王敏可,王甜甜,黄金标,林优优,谢继承. 全脑CT灌注联合深度学习全模型迭代算法用于评估急性缺血性脑卒中. 中国医学影像技术. 2025(04): 515-519 . 百度学术
5. 胡永志,武鹏飞,曹达,王彬彬,唐玉霞,王守巨,王传兵. 低管电压结合AIIR抑制髋关节假体CT伪影的效果研究. CT理论与应用研究(中英文). 2025(03): 377-384 . 百度学术
6. 张宝平,李傲,李宇航,朱书萌,田倩,赵文哲,肖瑶,侯伟,刘哲,王睿,黄欣,郝辉,王怡名,杨健,金超. 人工智能迭代重建算法对超低剂量胸部CT图像质量和计算机辅助肺结节检测的影响. 遵义医科大学学报. 2024(04): 384-391 . 百度学术
7. 中国医师协会放射医师分会,中华医学会影像技术分会,中国医师协会医学技师专业委员会,中国医学装备协会放射影像装备分会. CT深度学习图像重建算法临床应用中国专家共识. 中华放射学杂志. 2024(07): 697-706 . 百度学术
8. 曹建新 ,朱卓 ,陈颖 ,李嘉帅 ,张清 . 联合人工智能模型迭代重建和80 kV管电压降低腹部增强CT辐射剂量. 放射学实践. 2024(10): 1381-1386 . 百度学术
9. 马钦,龚欢,郭瑞,彭睿,宋法亮,马静. 人工智能迭代重组在病毒性肺炎诊断中的应用. 影像诊断与介入放射学. 2023(05): 342-346 . 百度学术
其他类型引用(2)