Influence Of Chest CT Combined With Various Reconstruction Algorithms On The Measurement And Display Of Pulmonary Nodules:A Phantom Research

Part One Effects of ASIR-V Standard Convolution Kernel on Measurement and Display of Pulmonary Nodules under Different Effective Doses（Conventional Dose and Low Dose）Objective:To explore the value of adaptive statistical iterative reconstruction-veo standard kernel（AV STND）on the measurement accuracy and the display of pulmonary nodules according to chest conventional dose and low dose CT.Methods:The conventional-dose(100k Vp,Smart m A,CTDI_VOI=3.27m Gy)and low-dose scan(100k Vp,40 m A,CTDI_VOI=0.84m Gy)were performed on the chest phantom with solid and ground-glass nodules of 5,8,and 10 mm in diameter.The raw data was used to reconstruct the chest CT images with lung kernel（AV LUNG）and standard kernel of ASIR-V（AV STND）under lung window.Measure and calculate the image noise and nodule signal-to-noise ratio（SNR）and contrast-to-noise ratio（CNR）,and calculate the mean CT value deviation and volume deviation of nodules against the nodule standard value.Then the image noise reduction,bronchovascular bundle sharpness,and nodule visibility were subjectively evaluated.Results:1.Compared with AV LUNG,AV STND has lower noise,higher SNR and CNR of solid and ground-glass nodules,higher image noise reduction score,and the sharpness of bronchovascular bundles was higher by AV LUNG（all P<0.05）according to chest conventional and low dose.2.The average CT value deviations of solid nodules were smaller at AV STND images（all P<0.001）,and the average CT value deviations of ground-glass nodules were smaller at AV LUNG images（all P<0.001）;The volume deviations of solid nodules and ground-glass nodules were smaller according to the AV LUNG image according to chest conventional and low dose（all P<0.05）.3.There was no significant difference in the visibility score of ground-glass nodules between two kinds of reconstructed image according to chest conventional dose（all P>0.05）,and the visibility score of ground-glass nodules reconstructed by AV STND according to low dose was higher（all P<0.001）;With respect to the visibility scores of solid nodules,there was no significant difference between the two convolution kernels of ASIR-V according to chest conventional and low dose（all P>0.05）.Conclusions:1.Compared with AV LUNG,the AV STND algorithm reduces the image noise and improves the SNR and CNR of lung nodules under the chest conventional dose and low dose.2.Under the chest conventional dose and low dose,AV STND reconstructed images were more accurate in measuring the mean CT value of solid nodules in the chest phantom;AV LUNG reconstructed images were more accurate in measuring all nodules’volume and mean CT value of ground-glass nodules in the chest phantom.3.Under the conventional dose,the visibility of pulmonary nodules according to the two convolution kernels is equivalent;under the low dose,the visibility of ground glass nodules according to the AV STND algorithm is better.Part Two Influence of Chest Low-Dose CT Combined with Deep Learning Reconstruction Algorithm on Measurement and Display of Pulmonary NodulesObjective:To investigate the effectivenss of chest low-dose CT combined with True Fidelity ^TMDeep-learning image reconstruction（DLIR）for the measurement and display of pulmonary nodules.Methods:The chest phantom with solid and ground-glass nodules of 5,8,and 10 mm in diameter underwent low dose scan(100k Vp,40 m A,CTDI_VOI=0.84 m Gy),and the raw data was reconstructed with adaptive statistical iterative reconstruction-veo（ASIR-V）and the M-strength and H-strength of DLIR（DLIR-M and DLIR-H）at standard kernel for image reconstruction under lung window.Measure and calculate the noise and SNR of lung tissue,and the noise,SNR and CNR of the nodules in the three groups of reconstructed images,and calculate the mean CT value deviation and volume deviation of the nodules against the standard value of the nodules.The degree of noise reduction,the sharpness of bronchovascular bundles,and the visibility of nodules were subjectively evaluated in the reconstructed images.Results:1.In the three groups of images,the noise of lung tissue and nodules with DLIR-H were the lowest,the SNR of lung tissue and nodules,and the CNR of nodules were the highest（all P<0.05）.The score of image noise reduction with DLIR-H was the highest（P<0.001）,and there wasn’t significant difference in the sharpness score of bronchovascular bundles（P=0.365）.2.There was no significant difference in the mean CT value’s deviation of the solid nodules measured in the three groups of reconstructed images（all P>0.05）,the mean CT value’s deviation of 10mm and 5mm ground-glass nodules were the less biased in DLIR-M、DLIR-H than ASIR-V images（all P<0.001）;There was not significantly difference in the volume’s deviation of 10mm and 8mm nodules according to various reconstructed images（all P>0.05）,and the volume’s deviation of 5 mm solid nodules in ASIR-V images was the smallest（P=0.020）,the volume of 5mm ground-glass nodule was the less biased in DLIR-M、DLIR-H images than ASIR-V（all P<0.05）.4.There was no significant difference in the visibility score of solid nodules and ground-glass nodules among the three groups of images（all P>0.05）.Conclusions:1.The DLIR-H algorithm reduces the noise of the chest phantom image,and improves the SNR and CNR of pulmonary nodules according to low dose.2.The measurement and display of nodules in DLIR images are comparable to those of ASIR-V images at low dose,and the measurement of5mm ground-glass nodule in DLIR image is more accurate especially.