SAT0508 A New Method of Quantitative Analysis of Thoracic CT Images to Investigate Connective Tissue Disease (CTD)-Associated Interstitial Pneumonia

Background Although the advent of high-resolution CT (HRCT) and multidetector CT (MDCT) have led to advances in the CT diagnosis of diffuse lung disease, this still mainly consists of macroscopic scoring. Recently, workstations have been developed that are capable of image processing using quantitative analysis methods based on MDCT, and the significance and clinical application are currently a matter of debate. Objectives Interstitial pneumonia (IP) is common in CTD, especially polymyositis (PM) and dermatomyositis (DM), and it frequently has a major effect on prognosis. In this study, a novel method of quantitative analysis of CT images (AZE Virtual Place) was applied for the quantitative diagnosis of IP in PM/DM patients. Methods The subjects were 20 myositis patients (11 PM, 9 DM) who underwent AZE Virtual Place analysis between January 2008 and June 2010. The entire lung field volume was calculated based on volume data of thoracic CT images for each patient. The relative volume of each CT value range was defined as the integrated pixel volumes in each range of CT values between every 100 Hounsfield units(HU) for CT values between –700 and +200 HU, divided by the total volume. We investigated each relative volume in patients with and without IP, and the correlations between the each relative volumes and labolatory parameters (Sialylated carbohydrate antigen KL-6[KL-6], surfactant protein-D [SPD], creatine kinase [CK], erythrocyte sedimentation rate [ESR], albumin [Alb], lymphocytes [Lym], and immunoglobulin G[IgG]). Mann-Whitney/Spearman’s test and the ROC curve were used for statistical analysis. Results Relative lung volumes were significantly higher in all CT value ranges examined in IP patients than in patients without IP. The AUC for IP was ³0.8 for all CT value ranges except from 0 to +200 HU (p < 0.01). ROC curve analysis revealed that between labolatory parameters and IP, IgG was significantly higher in IP patients, and Alb was significantly lower. KL-6, the conventional IP diagnostic marker, showed a particularly strong correlation with the relative volumes of the CT value range from –700 to –300 HU (ρ = 0.65, p< 0.01). ESR and IgG showed a positive correlation with all CT value ranges (ESR: ρ = 0.86, p < 0.01; IgG: ρ = 055, p < 0.01), whereas Alb showed a negative correlation (ρ= –0.66, p < 0.01). Conclusions Image processing of thoracic CT scan was carried out using AZE Virtual Place, a novel quantitative analysis method, in patients with my