探究全模型迭代算法对胰腺CT图像质量的改善情况

Explore the Improvement of Pancreatic CT Image Quality by FIRST

目的探索全模型迭代算法(Forward Projected Model-Based Iterative Reconstruction Solution,FIRST)对胰腺CT图像质量改善的价值。方法回顾性收集行腹部增强检查并且诊断胰腺正常的患者37例。所有患者均进行了平扫期、动脉期、实质期和延迟期四期扫描,对四期原始数据分别重建三组图像,分别为FIRST图像、混合迭代算法(Adaptive Iterative Dose Reduction 3D,AIDR 3D)图像和滤波反投影算法(Filtered Back Projection,FBP)图像。测量并计算图像的平均CT值、噪声指数SD值、信噪比(Signal-to-Noise Ratio,SNR)值和对比噪声比(Contrast to Noise Ratio,CNR)值等客观指标,并对三组图像分别进行主观评分统计。从客观和主观两方面来评估FIRST对胰腺CT图像质量的改善效果。结果客观指标的评价结果显示三种重建算法的四期图像CT值无统计学差异(P值均>0.05);SD值、SNR值和CNR值均具有统计学差异(P值均<0.001)。主观评分的结果显示FBP重建图像平扫期和延迟期不能满足诊断需要,其余图像均符合诊断需要,整体评分AIDR3D高于FIRST和FBP。结论FIRST技术相比较AIDR3D和FBP重建算法可以显著降低图像噪声。在满足临床诊断需求的情况下,FIRST技术在胰腺图像质量的客观评估中更具优势。

Objective To explore the value of forward projected model-based iterative reconstruction solution(FIRST)to improve the quality of pancreatic CT images.Methods All patients underwent four-phase scans in the plain,arterial,parenchymal,and delayed phases.Three sets of images were reconstructed from the raw data of the four phases,which were FIRST,adaptive iterative dose reduction 3D(AIDR 3D),and filtered back projection(FBP)images.The image’s average CT value,noise index SD value,signalto-noise ratio(SNR)value and contrast noise ratio(CNR)value and other objective indicators were measured and calculated,the subjective score statistics of the three groups of images were performed respectively.The improvement effect of FIRST on pancreatic CT image quality from both objective and subjective aspects were evaluated.Results The evaluation results of objective indicators showed that the CT values of the four phases of the three reconstruction algorithms were not statistically different(all P>0.05);SD,SNR and CNR values were all statistically different(all P<0.001).The subjective scoring results showed that the plain scan and delayed phases of FBP reconstructed images could not meet the diagnostic needs,and the rest of the images were in line with the diagnostic needs.The overall score AIDR 3D was higher than FIRST and FBP.Conclusion FIRST technology can significantly reduce image noise compared with AIDR 3D and FBP reconstruction algorithms.In meeting the needs of clinical diagnosis,FIRST technology has more advantages in the objective assessment of pancreatic image quality.