Objective:To investigate the impact of different CT reconstruction kernels on the quantitative analysis of small pulmonary vessels in patients with chronic obstructive pulmonary disease (COPD) and high-risk patients.Methods:This study was a cross-sectional study. Clinical and imaging data of 73 COPD and high-risk patients visiting the China-Japan Friendship Hospital between March and April 2024 were retrospectively analyzed. All patients underwent high-resolution CT of the chest and pulmonary function tests, with the ratio of forced expiratory volume in one second to forced vital capacity (FEV 1/FVC) obtained. The raw CT data were reconstructed using different kernels: the FC86 group used the adaptive iterative dose reduction(AIDR) 3D standard lung sharp reconstruction algorithm, the FC18 group used the AIDR 3D standard Body standard reconstruction algorithm, the advanced intelligent clear-IQ engine(AiCE) Lung group used the AiCE deep learning reconstruction algorithm for lung, and the AiCE Body group used the AiCE deep learning reconstruction algorithm for body. Image signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR) and image noise were calculated. The pulmonary vessel segmentation & Measurement software was used to segment and extract pulmonary arteries and veins of four groups in thin-slice whole-lung CT imaging, obtaining the ratio of small pulmonary arteries (%V artery<5) and the ratio of small pulmonary veins (%V vein<5). The One-way repeated measures ANOVA or Friedman test was used to compare the differences in SNR, CNR, image noise, %V artery<5, and %V vein<5 among the four groups, followed by Bonferroni post hoc or Bonferroni-Dunn test with P-value correction to analysis differences between subgroups. The correlations between %V artery<5 and FEV 1/FVC, as well as between %V vein<5 and FEV 1/FVC were analyzed using Spearman rank correlation analysis in all four groups. Results:The overall differences in image noise, SNR, and CNR in the AiCE Lung, AiCE Body, FC18, and FC86 groups were statistically significant ( P<0.001). Except for the difference in CNR values between the AiCE Lung group and the FC18 group, which was not statistically significant ( P=0.192), all differences were statistically significant ( P<0.016 7). The overall differences in %V artery<5 values and %V vein<5 values in the AiCE Lung, AiCE Body, FC18, and FC86 groups were statistically significant ( P<0.001). The %V artery<5 and %V vein<5 values in the FC18 group were lower than those in the AiCE Lung, AiCE Body, and FC86 groups ( P<0.016 7), and the rest of the differences were not statistically significant ( P>0.016 7). %V artery<5 and %V vein<5 were positively correlated with FEV 1/FVC in all 4 groups ( P<0.05), with the highest correlation coefficient between %V vein<5 and FEV 1/FVC in the AiCE Body group ( r=0.501, P=0.001). Conclusions:DLR-AiCE-based kernel reconstruction optimizes image quality and significantly affects the results of quantitative parameters of small pulmonary vessels. The reconstruction kernel prioritized for quantitative analysis of small vessels within the lungs in COPD based on the CT scanner in this study is AiCE Body.

Objective:To investigate the impact of different CT reconstruction kernels on the quantitative analysis of small pulmonary vessels in patients with chronic obstructive pulmonary disease (COPD) and high-risk patients.Methods:This study was a cross-sectional study. Clinical and imaging data of 73 COPD and high-risk patients visiting the China-Japan Friendship Hospital between March and April 2024 were retrospectively analyzed. All patients underwent high-resolution CT of the chest and pulmonary function tests, with the ratio of forced expiratory volume in one second to forced vital capacity (FEV 1/FVC) obtained. The raw CT data were reconstructed using different kernels: the FC86 group used the adaptive iterative dose reduction(AIDR) 3D standard lung sharp reconstruction algorithm, the FC18 group used the AIDR 3D standard Body standard reconstruction algorithm, the advanced intelligent clear-IQ engine(AiCE) Lung group used the AiCE deep learning reconstruction algorithm for lung, and the AiCE Body group used the AiCE deep learning reconstruction algorithm for body. Image signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR) and image noise were calculated. The pulmonary vessel segmentation & Measurement software was used to segment and extract pulmonary arteries and veins of four groups in thin-slice whole-lung CT imaging, obtaining the ratio of small pulmonary arteries (%V artery<5) and the ratio of small pulmonary veins (%V vein<5). The One-way repeated measures ANOVA or Friedman test was used to compare the differences in SNR, CNR, image noise, %V artery<5, and %V vein<5 among the four groups, followed by Bonferroni post hoc or Bonferroni-Dunn test with P-value correction to analysis differences between subgroups. The correlations between %V artery<5 and FEV 1/FVC, as well as between %V vein<5 and FEV 1/FVC were analyzed using Spearman rank correlation analysis in all four groups. Results:The overall differences in image noise, SNR, and CNR in the AiCE Lung, AiCE Body, FC18, and FC86 groups were statistically significant ( P<0.001). Except for the difference in CNR values between the AiCE Lung group and the FC18 group, which was not statistically significant ( P=0.192), all differences were statistically significant ( P<0.016 7). The overall differences in %V artery<5 values and %V vein<5 values in the AiCE Lung, AiCE Body, FC18, and FC86 groups were statistically significant ( P<0.001). The %V artery<5 and %V vein<5 values in the FC18 group were lower than those in the AiCE Lung, AiCE Body, and FC86 groups ( P<0.016 7), and the rest of the differences were not statistically significant ( P>0.016 7). %V artery<5 and %V vein<5 were positively correlated with FEV 1/FVC in all 4 groups ( P<0.05), with the highest correlation coefficient between %V vein<5 and FEV 1/FVC in the AiCE Body group ( r=0.501, P=0.001). Conclusions:DLR-AiCE-based kernel reconstruction optimizes image quality and significantly affects the results of quantitative parameters of small pulmonary vessels. The reconstruction kernel prioritized for quantitative analysis of small vessels within the lungs in COPD based on the CT scanner in this study is AiCE Body.