Qualitative And Quantitative Assessment Clinical Application Of Chronic Obstructive Pulmonary Disease On HRCT

Part 1 Quantitative Assessment Clinical Application of Chronic Obstructive Pulmonary Disease On HRCTPurpose:The low attenuation areas volume percentage(LAA%)of individual pulmonary lobe can be measured automatically by High-resolution CT volume quantitative techniques.The obtained quantitative parameters will have a good comparability study with the index of pulmonary function test(PFT),and then the damage degree of pulmonary function in patients can be assessed synthetically.Thus,this technique can effectively provide valuable images for the diagnosis and treatment scheme in clinic.Materials and methods:This study included 83 COPD patients who underwent HRCT and also did the pulmonary function test in one week from December 2015 to December 2016.For each patient,images were acquired by 64-slice CT at full inspiration and then original data were transferred to post-processing workstation GEadw4.6.Gas trapping areas were systematically defined as percent lung tissue<-950HU.The post-processing workstation of parenchyma analysis worked out the parameters as follows:Total emphysema volume(TEV),Total Lung Volume(TLV),total LAA%,right lung LAA%,left lung LAA%,right upper lobe LAA%,right middle lobe LAA%,right lower lobe LAA%,left supper lobe LAA%,left lower lobe LAA%.The physiological measurements of pulmonary function parameters in this article as follows:forced expiratory volume in one second(FEV1),FEV1%predicted,forced vital capacity(FVC),FEV1,FEV1/FVC,peak expiratory flow(PEF),forced expiratory flow at 25%of FVC exhaled(FEF25),FEF50,FEF75,diffusion capacity of carbon monoxide in the lung predicted(DLCO%),the ratio of residual volume and total lung(RV/TLC).Spearman's rank correlation coefficients were calculated between the quantitative CT values and aforementioned parameters.CT values between COPD subtypes was anylized by One-Way ANOVA,and comparison among groups was used LSD analysis.Kruskal-Wallis text analyzed the difference between COPD subtype and FEV1/FVC,FEV1%and LAA%.P values less than 0.05 were considered different;P values less than 0.01 were considered statistically significant different.All statistical analyses were done by using SPSS version 20.0.Results:The results showed that the body mass index(BMI)were ranged from 13.3 to 28.1 of all 83 male patients(47?85 years old).Their mean age was 66 years old and twelve patients among them were without the experience of smoking.According to the last classification of COPD stages,8 cases were GOLD 1,33 cases were GOLD2,27 cases were GOLD3,15 cases were GOLD4 in this study.Quantitative CT values including total lung LAA%,left lower lobe LAA%,left lung LAA%,right lobe LAA%,right lower lobe LAA%showed statistical correlation with FEV1,FEV1%predicted,FEV1/FVC,FEF25,FEF50.Bilateral lower lung LAA%found correlation with PEF75,which indicated significant difference with physiological measurements of pulmonary function(FEV1,FEV1%predicted,FEV1/FVC,PEF).Correlation between FEV1/FVC and TEV was significant(r=0.759,P<0.001).And correlation between TLV and TLC was significant(r=-0.355,P=0.001).However,correlation between DLCO%and Bilateral upper lung LAA%was significant(r=-0.473,P=0.026).Bilateral upper lung LAA%of GOLD 1 cases displayed statistical significant difference with GOLD3.GOLD 1 and GOLD4 in left lower lobe LAA%had also significant difference.Besides right middle lobe LAA%,the remaining lobes showed difference between GOLD2 and GOLD3.Furthermore,GOLD2 and GOLD4 showed also significant difference in left lower lobe LAA%and lung LAA%.Conclusions:In summary,Bilateral lower lung LAA%had significant correlation with physiological measurements of pulmonary function(FEV1,FEV1%predicted,FEV1/FVC,FEF50,PEF),but had no obvious correlation with RV/TLC and DLCO%.Meanwhile,DLCO%and Bilateral upper lung LAA%showed slight correlation significance.Therefore,LAA%in each lobe could reflect injury lesions and the severity,which may provide evidence to tailor personalized treatments.Part 2 Qualitative Assessment of Emphysema Subtype By Computer Post-Processing Technology on HRCTPurpose:Computer Post-processing technology provides a repeatable,unbiased,accurate method for automatic identification of emphysema subtypes,and this technology have the ability to identify different morphologies and degree of pathological changes of emphysema.Thus it is able to provide a new idea for clinician when they plan personalized treatment programs for COPD patients.Materials and methods:For 6 emphysema patients and 6 normal patients,images were acquired by 64-slice CT at full inspiration and then original data was transferred to post-processing ITK-SNAP software.The different colors represented different subtypes of emphysema and normal lung tissue;Normal tissue(NT)marked red,Centrilobular Emphysema(CLE)marked green label,Panlobular Emphysema(PLE)marked blue mark,Paraseptal Emphysema(PSE)mark yellow.1000 different abnormal tissue ROI(Region of interest)was randomly chose in each 6 marked cases(Emphysema Group),and these ROI had a corresponding label,including the CLE,PLE,PSE.A total of 6000 abnormal tissue ROI was included.Besides,another 1000 NT of ROI was selected in emphysema group.A random sample of 1000 NT ROI in normal group.1000 ROI was randomly selected from each type of emphysema in 6000 abnormal ROI subtypes(CLE,PSE,PLE).Then 1000 ROI of the normal group as the training samples,the remaining 3000 abnormal ROI and 1000 normal tissues ROI in the emphysema group was test samples.200 ROI was randomly selected as test samples in each subtype of emphysema(CLE,PSE,PLE)and NT.Intensity(INT),the Rotation invariant LBPs(RILBPs),INT + RILBPs was applied to automatic identify of emphysema subtype respectively by using computer post-processing.This experiment repeated 5 times,average values reflected the accuracy of the classification.Results:INT,RILBPs,INT + RILBPs used computer post-processing automatic to identify emphysema subtype classification and reflected the accuracy of the classification;CLE,PLE,PSE of three emphysema subtypes were tested,the accuracy as follows:88.28%in INT method,86.46%in RILBPs,and 94.62%in INT + RILBPs.While NT was added in this calculation,the classification accuracy was declined;INT method was 72.09%,RILBPs method was 67.34%,the INT+ RILBPs was 84.29%.Thus INT+RILBPs method possessed the highest classification accuracy.Conclusions:INT,RILBPs,INT + RILBPs used computer post-processing automatic to identify emphysema subtype classification and reflected the accuracy of the classification;Classification accuracy of INT+RILBPs method was higher than INT or RILBPs.The accuracy of the classification was decreased when NT was added.INT + RILBPs method provides a new diagnosis assessment for automatic identification emphysema subtype in clinical.