Auto-kV技术联合AI深度学习算法(Clear Infinity)在冠脉“精准”双低成像中的应用  

Applications of Auto-kV technique combined with AI deep learning algorithm(Clear Infinity)in coronary CTA with low dose and low iodine load

作　　者：张斌[1] 高思喆[1] 郭文力[1] 李雨桐 ZHANG Bin;GAO Si-zhe;GUO Wen-li;LI Yu-tong(Department of Radiology,Shengjing Hospital of China Medical University,Shenyang 110004,China;CT Products Business Division,Neusoft Medical System Co.,Ltd,Shenyang 110167,China)

机构地区：[1]中国医科大学附属盛京医院放射科,辽宁沈阳110004 [2]东软医疗系统股份有限公司CT产品事业部,辽宁沈阳110167

出　　处：《中国临床医学影像杂志》2024年第4期283-287,共5页Journal of China Clinic Medical Imaging

基　　金：辽宁省高性能医疗器械应用示范项目[202001]。

摘　　要：目的:探讨Auto-kV技术联合AI深度学习算法(Clear Infinity)在冠状动脉“双低”成像中的价值。方法:连续选取62位行冠脉CTA检查者,分为A、B两组,每组31例。A组(实验组)采用Auto-kV技术扫描,O-Dose=0.7;B组(对照组)采用常规120kV扫描,O-Dose=1.0。A组对比剂用量依据Auto-kV技术给定管电压:140kV-0.9mL/kg;120kV-0.8mL/kg;100kV-0.7mL/kg;80 kV-0.6 mL/kg;70 kV-0.5 mL/kg;60 kV-0.4 mL/kg,B组对比剂用量统一为0.8 mL/kg。对比剂注射速率=对比剂用量/12 s。将A组原始数据采用迭代重建算法(Clear View)及AI重建算法(Clear Infinity)重建出两组图像,分别命名为A1组和A2组。迭代及AI算法的权重均设定为70%。B组原始数据采用滤波反投影法(FBP)重建。采用统计学分析图像质量、碘负荷及辐射剂量的差异。结果:A1、A2组图像冠脉各节段内CT值均高于B组,差异均有统计学意义(P均<0.001)。A1、A2组的噪声显著低于B组,差异有统计学意义(P均<0.001)。A1、A2组的信噪比及对比噪声比显著高于B组,差异有统计学意义(P均<0.001)。A、B两组间CTDI、DLP、ED及碘负荷的差异均有统计学意义(P均<0.001),A组的碘负荷及辐射剂量均显著低于B组。主观评测中三组间对比度无统计学差异;A1、A2组的主观噪声及诊断置信度优于B组,有统计学意义,A1、A2组间没有统计学意义;在锐利度及细节显示上三组间均有统计学意义,且A2组优于A1组优于B组。结论:Auto-kV技术联合AI深度学习算法(Clear Infinity)应用于冠状动脉CTA中可实现个体化扫描,显著降低对比剂用量及辐射剂量,并保证图像质量。Objective:To explore the value of Auto-kV technology combined with AI deep learning algorithm(Clear Infinity)in coronary CTA with low dose and low iodine load.Methods:Sixty-two patients who underwent coronary CTA were selected and divided into two groups of 31 cases each:group A(experimental group)was scanned by Auto-kV technology with O-Dose=0.7;group B(control group)was scanned by conventional 120 kV with O-Dose=1.0.The dosage of contrast agent of group A was based on the tube voltages given by Auto-kV technology:140 kV-0.9 mL/kg,120 kV-0.8 mL/kg,100 kV-0.7 mL/kg,80 kV-0.6 mL/kg,70 kV-0.5 mL/kg,60 kV-0.4 mL/kg,and the dosage of group B was standardized as 0.8 mL/kg.The injection time was 12 seconds.The raw data of group A were reconstructed by iterative reconstruction algorithm(Clear View)and AI reconstruction algorithm(Clear Infinity)to produce two groups of images,named group Al and group A2,respectively.The weights of both the iterative and AI algorithms were set to 70%.The raw data of group B were reconstructed by filtered back projection(FBP).The differences in image quality,iodine load and radiation dose were analyzed using statistics.Results:There were significant differences in CT value among the three groups,the CT value of group Al and A2 were higher than group B.There were significant differences in image noise among the three groups,the image noise of group Al and A2 was significantly lower than group B.There were significant differences in SNR and CNR among the three groups.The SNR and CNR of group A1 and A2 were significantly higher than group B.There were significant differences in CTDI,DLD,ED and iodine load between groups A and B.There was no statistical difference in subjective contrast among the three groups.The subjective noise and diagnostic confidence of group Al and A2 were better than group B,not between groups Al and A2.There were significant differences in sharpness and detail among the three groups.Group A2 was better than group Al and better than group B.Conclusion:Auto-kV technology

关 键 词：冠状血管 体层摄影术 X线计算机 

分 类 号：R543.3[医药卫生—心血管疾病] R814.42[医药卫生—内科学]