Applications Of Iterative Reconstruction Algorithm In Routine Chest CT Examinations And Image Quality Assessment

Part Onethe Best Weight Range of Adaptive Statistical Iterative Reconstruction Apply toRoutine Dose Chest CTObjective: To investigate the best weight range of adaptive statistical iterativereconstruction（ASiR） applied to routine dose chest CT Methods: With institutionalreview board approval,22adult patients who underwent an enhanced chest CTexamination were enrolled. Scanning parameters included a pitch of0.984:1,100kilovolts (peak), noise index15, auto current,40mm table feed per rotation. Raw datewere reconstructed with different weightings of ASiR(10-100%,step10%)respectively,and the reconstructed thickness was0.625mm.Image noises were measured andcontrast-to-noise ratio (CNR) of thoracic aorta relative to back muscle was assessed.Image quality was assessed visually with a preset lung window settings[window width(WW),1500Hounsfield units(HU),window level(WL),-600HU] and mediastinalwindow settings(WW,400HU;WL,40HU) using a6-point scale. The correlationbetween image noise and different weight of ASiR,contrast-to-noise ratio and differentweigh of ASiR using Pearson’s correlation analysis for statistical analysis. Results: Theimage noise reduced while the weight of ASiR increasing, there was a significantnegative correlation value between The image noisen and the weight of ASiR(r=0.997,P <0.001).While the weight of ASiR setting100%,the image noise was15.56±3.02HUwhich was minimum of all. the image noise could be reduced by48.08%than10%ASiR(29.64±3.70HU）.And contrast-to-noise ratio and different weigh of ASiR showeda significant positive correlation (r=0.986, P <0.001). While the weight of ASiRsetting10%, the CNR was8.60±2.88which was the minimum of all.The CNR was thelargest while the weight of ASiR setting100%.The mean lung window subjectivescores of image quality were always higher than the mediastinum window’s.Whilethe weight of ASiR setting40-60%,the mage noise was24.30±3.55HU、17.11±2.55HUand11.69±1.74HU,respectively.And the mean lung and mediastinum windowsubjective scores of image quality were the highest that were more than5points.In60%ASiR later, the mean lung window and mediastinum subjective scores of image qualitydropped, which were less than4points. Conclusion:In the routine doses, take intoaccount the mean lung window and mediastinum subjective scores of image quality,while the weight of ASiR setting40%to60%, the obtained image quality noise andcontrast-to-noise ratio are higher, and the mean lung window and mediastinumsubjective scores of image quality are the highest of all. Part twoImpact of Reconstruction Algorithm in Routine DoseChest CT Image Quality: Comparison of FBP, ASiR and VEOObjective:To investigate the impact of different reconstruction algorithms, includingfiltered back projection(FBP), adaptive statistical iterative reconstruction(ASiR) andmodel-based iterative reconstruction(MBIR, with VEO as its trade name) in imagequality of the routine dose chest Computed Tomography.Methods: With institutionalreview board approval,37adult patients who underwent an enhanced chest CT wereenrolled. Scanning parameters included a pitch of0.984:1,100kilovolts (peak), noiseindex15, auto current,40-mm table feed per rotation. Raw data was reconstructed withFBP,50%ASiR (blending of50%ASiR and50%FBP image) and VEO algorithmrespectively; the reconstructed section thickness was0.625mm.Image noise wasmeasured and contrast-to-noise ratio (CNR) of thoracic aorta relative to back musclewere assessed. Image quality was assessed visually with a preset lung windowsettings[window width (WW),1500Hounsfield units(HU),window level(WL),-600HU]and mediastinal window settings(WW,400HU;WL,40HU) using a6-point scale.Results:The image noise of FBP，50%ASIR and VEO was28.74±4.90HU,20.81±4.18HU and13.16±2.48HU,respectively.Compared with FBP, objective image noisereduced by54.21%(P=0.000)and27.59%(P=0.000) in images reconstructed with VEOand50%ASiR,respectively.The CNR of thoracic aorta to back muscle forFBP,50%ASiR and VEO was9.05±3.01,12.81±4.68and19.54±6.10，respectively．Compared with FBP，CNR of images reconstructed with50%ASiR andVEO increased by41.55%(P=0.000) and115.91%(P=0.000)，respectively．The meanlung window subjective score of image quality reconstructed with FBP，50%ASIR andVEO was3.84±0.42、4.98±0.31and5.76±0.19,respectively. Compared with FBP,objective image noise reduced by50.00%(P=0.000) and29.69%(P=0.001) inimages reconstructed with VEO and50％ASiR,respectively.The mean mediastinalwindow subjective score of image quality reconstructed with VEO was5.55±0.16,higher than that of FBP (3.84±0.42,P=0.000) and50%ASiR (4.98±0.31,P=0.001).Conclusion:VEO and ASiR reconstruction techniques have the ability to reduce imagenoise and improve the image quality compared with the current algorithms such as FBP,especially VEO technique.