| Purpose:(1)To investigate the effect of different reconstruction matrices of CT on pulmonary nodules measurement and chest image quality based on anthropomorphic chest phantom.(2)A large reconstruction matrix of 1024×1024 combined with Karl iterative reconstruction algorithm was used to perform low-dose lung CT imaging,and its clinical application value was explored through the evaluation of small pulmonary nodules and signs display.Materials and methods:(1)Phantom study: A total of 15 simulation of spherical nodules of different sizes(3 mm,5 mm,8 mm,10 mm,12 mm)and densities(-630 HU,-800 HU,+100 HU)were randomly placed in a PH-1 anthropomorphic chest phantom(Kyoto Kagaku).The phantom was scanned using the u CT760 and reconstructed using three different matrices(512×512,768×768,1024×1024)based on the scanned raw scan data to obtain three sets of A,B,and C images.Spherical nodules were analyzed on a postprocessing workstation using computer-aided diagnosis software(CAD)for pulmonary nodules,and the maximum cross-sectional long and short diameters of nodules were measured on axial images,and the sum of long and short diameter errors,out-ofroundness and absolute error rate(APE)of volume measurements were calculated.CT and SD values were measured at the apical lung,the largest level of the thorax,the homogeneous region of the lung base and the parasternal soft tissue at the level of the tracheal ridge,the aortic arch and the largest level of the heart.Image quality and pulmonary nodule display under different matrices were evaluated by two observers using a double-blind method on a 5-point scale.One-way ANOVA was used to compare the differences between the error sum of long and short diameter of pulmonary nodules,image CT values,and SD values in groups A,B,and C.The agreement between the subjective scores of images by two radiologists was tested by Kappa test,and the out-ofroundness of nodules,volume APE and subjective quality scores of images were tested by Friedman test.(2)Clinical Study: Totally 500 patients who underwent chest CT examination in our hospital were collected,and all of them were scanned by low-dose chest CT using u CT760(United Imaging Corporation).After the scans were completed,the reconstruction was performed based on the original scan data.group A used the conventional 512×512 matrix combined with Karl level 5 reconstruction;group B used the 1024×1024 large reconstruction matrix combined with different levels of Karl algorithm to obtain four subgroups of B1(Karl 6),B2(Karl 7),B3(Karl 8)and B4(Karl9).The remaining reconstruction parameters of the two groups are the same,with reconstruction layer thickness of 1 mm,layer interval of 1 mm,and filter function of bone reconstruction(B＿SHARP＿C).The reconstructed thin-layer images of groups A and B were analyzed,and the CT and SD values of the air in the tracheal lumen above the aortic arch(background)and the homogeneous lung parenchyma in the upper lobe of the left lung were measured,and the SNR of the images was calculated,and the lung image quality of groups A and B was evaluated by two physicians using an independent doubleblind method on a 5-point scale.The subgroup with the best image quality in group B was obtained by combining objective and subjective image quality evaluation,and compared with group A.The nodule detection ability,display of pulmonary nodules and morphological characteristics were compared between the two reconstruction conditions.The cases were followed up clinically and the diagnostic efficacy of the subgroup with the best image quality in groups A and B was compared for pulmonary nodules based on the pathological findings of the patients’ pulmonary nodules.Results:(1)The differences in pulmonary nodule length and short diameter error sum,out-of-roundness and APE between groups A to C were statistically significant(all P<0.05)and gradually decreased with increasing matrix.the differences in CT values between groups A to C were not statistically significant(P>0.05);SD values increased with increasing matrix and the differences were statistically significant(P<0.001).There was good agreement between the subjective scores of the two observers(Kappa=0.71-0.90).Group C had the highest overall subjective score.(2)The differences in CT values of lung parenchyma between groups A and B were not statistically significant(P>0.05),while the differences in background CT values,SD values of trachea and lung parenchyma,and SNR values of trachea and lung parenchyma were statistically significant(P<0.05).within group B,the SD values of trachea and lung parenchyma gradually decreased and the SNR values gradually increased with the increase of Karl grade(P<0.05).The subjective evaluation of image quality was in good agreement between the two radiologists for each group(Kappa=0.796-0.987,P<0.05),and all subjective scores were higher in group B than in group A(P<0.05),with the highest subjective scores in group B4.Statistical analysis of nodules in groups A and B4 showed that the number of pulmonary nodulesdetected was comparable;after comparing the nodule density classification,groups A and B4 showed a significant difference in the number of partially solid The difference in display clarity between group A and group B4 was not statistically significant(P>0.05)for some solid nodules(≤3 mm)and solid nodules(>6 mm),while the difference in display clarity for the rest of nodules of different density sizes was statistically significant(P<0.05);compared with group A,the display clarity of nodules in group B4 improved by 12%-100%.In the signs of tumor-pulmonary interface,burr sign,lobar sign,vacuole sign,air bronchus sign,and halo sign,group B4 was better than group A(P<0.05),and the difference in the display of pleural depression sign was not statistically significant(P>0.05).43 patients underwent surgical pathological examination,and the diagnostic accuracy rate of group B4 was 65.11%,which was better than that of group A by 38.88% using pathological results as the gold standard(P<0.05).Conclusion:(1)The large reconstruction matrix has higher accuracy in the measurement of pulmonary nodules and better display of pulmonary fine structure,which is helpful for the accurate diagnosis of pulmonary lesions.(2)Under LDCT,the combined application of large reconstruction matrix and Karl algorithm can obtain better image quality and achieve accurate diagnosis of pulmonary nodules. |
PDF Full Text Request