人工智能优化算法对提高大体型患者低剂量扫描冠状动脉图像质量的价值

Impact of artificial intelligence-based optimization algorithm on image quality of low dose coronary CT angiography in big size patients

目的:探讨人工智能(AI)图像优化算法对提高大体型患者低剂量扫描冠状动脉图像质量的价值。方法:前瞻性连续纳入2018年2-5月在本院NeuViz128CT行冠状动脉CTA检查的28例大体型患者(BMI>26kg/m^2)。所有的患者均采用步进扫描模式,管电压100kV,自动管电流调制(233.4±46.7mAs)。对原始数据采用迭代算法(Clearview+50%)进行重建得到A组图像,进一步对该组图像采用AI图像优化技术进行处理,所得图像作为B组。分别在主动脉根部、左主干开口、左前降支中段、左回旋支中段及右冠状动脉中段选取不同的兴趣区,测量这两组图像的冠状动脉的CT值、噪声、信噪比(SNR)、对比噪声比(CNR),由两名高年资的医生以Likert4级评分法评估该两组图像的主观评分(1分,优秀;4分,不能诊断)。结果:患者平均BMI为(29.31±3.19)kg/m^2,平均心率(64.89±8.13)次/分。与A组图像相比,B组图像主动脉根部、左主干开口、左前降支中段、左回旋支中段及右冠状动脉中段的噪声分别降低了68.36%、45.89%、28.41%、32.49%和31.25%。B组图像SNR和CNR明显优于A组图像(P均<0.01)。B组冠状动脉主观评分明显优于A组图像质量(1.66±0.27vs.1.82±0.20,P<0.001)。扫描过程中CT剂量指数为(10.6±0.9)mGy,剂量长度乘积为(167.8±26.2)mGy·cm,有效剂量为(2.3±0.4)mSv。结论:AI图像优化算法可以有效提高大体型患者在低剂量扫描时的冠状动脉图像质量,为大体型患者降低辐射剂量及优化冠脉动脉图像质量提供了新思路和新方法。

Objective: The aim of this study was to assess the impact of artificial intelligence (AI)-based optimization algorithm on image quality (IQ) with low dose coronary CT angiography (CCTA) in big size patients. Methods: Twenty-eight consecutive patients with a body mass index greater than 26kg/m^2 underwent clinical indicated CCTA examinations on NeuViz 128 CT scanner.All CCTA examinations were performed with step-and-shoot mode using 100kV,automatic current modulation (233.4±46.7mAs) and 60mL contrast medium.Images were reconstructed with iterative algorithm (group A) and further AI-based optimization algorithm (group B).Image noise,signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were calculated on the aorta root (Ao),left main artery (LM),left anterior descending (LAD),left circumflex (LCX),and right coronary artery (RCA) to evaluate IQ objectively.Subjective IQ score was graded independently and blindly by two radiologists with a four-point scale (1 for excellent and 4 for poor).In addition,CT dose index volumes (CTDIvol),dose length product (DLP) and effective dose (ED) were recorded. Results: The average body mass index (BMI) and heart rate were (29.31±3.19) kg/m^2,(64.89±8.13)min -1 respectively.Compared with group A,group B yielded significantly reduction in image noise in the Ao,LM,LAD,LCX and RCA of 68.36%,45.89%,28.41%,32.49% and 31.25%,respectively.SNR and CNR were both significantly increased in group B compared with group A (all P <0.01).The subjective IQ score has significantly difference between group A and group B (1.82±0.20 vs.1.66±0.27, P <0.001).The CTDIvol,DLP and ED was (10.6±0.9)mGy,(167.8±26.2)mGy·cm and (2.3±0.4)mSv. Conclusion: The AI-based optimization algorithm can effectively improve IQ of coronary CT angiography with low kV scan in big size patients,which can be an innovative and promising method.