RPS 804 - Quantitative analysis of chest CT findings

Author Block: M. Opitz, S. Zensen, J. Haubold, B. M. Schaarschmidt, M. Forsting, L. Umutlu, R. Hosch, F. Nensa, D. Westhölter; Essen/DE
Purpose: A poor nutritional status is associated with worse pulmonary function and survival in people with cystic fibrosis (pwCF). CF transmembrane conductance regulator (CFTR) modulators can improve both pulmonary function and body weight, but more data is needed to evaluate its effects on body composition. This study aims to investigate the body composition in pwCF receiving triple-combination elexacaftor/ tezacaftor/ ivacaftor (ETI) therapy.
Methods or Background: A pre-trained, deep-learning network was used to perform a fully automated body composition analysis (BCA) on chest CTs from adult pwCF before and after receiving ETI therapy. Muscle and adipose tissues were quantified and divided by bone volume to obtain body size-adjusted ratios. Results from BCA were correlated with pulmonary function parameters.
Results or Findings: At baseline, chest CT-based BCA was conducted in 66 pwCF, with 33 (50%) receiving either mono or dual-combination CFTR modulator therapy. Mono/ dual-combination CFTR modulator therapy was associated with higher intra- and intermuscular adipose tissue and epicardial adipose tissue ratios, while other BCA markers remained unchanged. After receiving triple-combination ETI, marked increases were observed in all adipose tissue ratios among pwCF, including the total adipose tissue ratio (+46.21%, p<0.001). In contrast, only small, but statistically significant, increases of the muscle ratio were measured in the overall study population (+1.63%, p=0.008). There were weak associations between the rate of change of the muscle ratio and the rate of change of percent predicted FEV1 (r=0.360, p=0.004).
Conclusion: Our findings suggest that CFTR modulator therapies primarily affect adipose tissues, not muscle tissue, in adults with CF. These findings may have implications for the future nutritional management of pwCF.
Limitations: CT-based BCA was performed in chest CT scans. Abdominal CT scans or full-body CT scans were not available and body composition might differ in other parts of the body.
Funding for this study: Funding was received from the German Research Foundation (DFG)-initiated clinician scientist program FU 356/12–2 (DW, JH).
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by the Institutional Review Board (no. 22-11073-BO).

Author Block: F. B. Pallasch, J. Weiß, F. Bamberg, M. Jung, J. B. Fingerhut, M. Reisert; Freiburg im Breisgau/DE
Purpose: There is increasing evidence that body composition is of prognostic relevance. While most studies focus on musculature, little is known about adipose tissue. The purpose of this study was to apply a deep-learning model for automatic 3D quantification of subcutaneous adipose tissue (SAT) on chest CT and investigate its association with mortality in a lung cancer screening population.
Methods or Background: 3D SAT was automatically segmented on chest CT of 26,144 individuals participating in the National Lung Screening Trial at baseline and after one year follow-up (n=52,228 scans). SAT volume (SATvol) and density (SATHU) were quantified. The primary outcome was all-cause mortality. Additional outcomes were lung cancer and cardiovascular mortality. Cox regression was used to assess the association between SATvol and SATHU at baseline, and a decrease in SATvol and SATHU (≥10%) and mortality.
Results or Findings: In 26,144 individuals (age 61.4±5.0 years; 40.9% female) 1839 (7%) deaths occurred over a median follow-up of 6.5 years. At baseline, only SATHU was associated with all-cause mortality after multivariable adjustment for clinical risk factors (age, sex, race, smoking status, pack years, prevalent hypertension, diabetes, past stroke and myocardial infarction (HR: 1.07 95%CI (1.02-1.12); p=0.003). After one year, individuals with a decline in SATvol or SATHU ≥10% had a significantly worse outcome compared to stable SATvol or SATHU (HR adjusted for the same risk factors SATvol: HR 1.94, 95%CI (1.77-2.13), p<0.001; SATHU: HR 3.1, 95%CI (2.79-3.44), p<0.001). Similar associations were found for lung cancer and cardiovascular mortality.
Conclusion: Deep learning models can quantify SAT on chest CT scans. SATHU at baseline and a decrease in SATvol and SATHU within one year are associated with mortality beyond clinical risk factors, which may help to improve personalised risk assessment.
Limitations: The study is limited by its retrospective analysis.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information: The study is based on a publicly available dataset.

Author Block: M. Giannetti, S. Paciulli, A. Gigante, A. Colalillo, L. Marchitelli, L. Conia, C. Catalano, N. Galea; Rome/IT
Purpose: Systemic sclerosis (SSc) is a multi-organ disease, which can affect the lung parenchyma resulting in an interstitial lung disease (ILD). Pulmonary function in those patients is influenced by lung disease and chest wall muscles (CWM) activity. High-resolution CT (HRCT) is a valuable tool to evaluate ILD progression and CWM atrophy. The purpose of this study was to investigate the potential contribution of CWMarea (CWMA) to the ventilatory efficiency and exercise capacity in patients with SSc.
Methods or Background: Forty-four SSc patients undergone cardiopulmonary exercise testing (CPET), HRCT and transthoracic echocardiography (TTE) were retrospectively enrolled. The CWMA was manually traced at the level of the 9th thoracic vertebra on CT images. CWMA was correlated with SSc duration and CPET parameters.
Results or Findings: The patient median age was 53 (IQR 43.5-58) with a BMI of 22.10 kg/m2 (IQR 24.25-20.55).
The median disease duration was 9.50 years (IQR 5.5-15) and the median CWMA was 43.9 cm^2 (IQR 36.8-56.5). We found a median %FVC of 100 (IQR 107-88), a %FEV1 of 95 (IQR 104.5-86), a %TLC of 93 (IQR 103.5-81.5), a %DLCO of 77 (IQR 86-68.5), and a DLCO/Va of 81.5% (IQR 93.5-69) for pulmonary functional parameters. TTE revealed a median EF of 60% (IQR 62.5-60), a RV and LV transverse diameter of 29mm (IQR 31-26 and 44mm (IQR 46-42) respectively), a TAPSE of 22 (IQR 25-21), and a PAPs of 28 (IQR 30-25). CPET parameters used were Watt-max 80% (IQR 105.5-62), VO2max 1211ml/min (IQR 1451-1026.5), VO2max 20.7ml/min/Kg (IQR23.82-17.9), VO2@AT 789.0ml/min (IQR 952-679), OUES 1364ml/min/L/min (IQR 1629-1229), VEmax 49.55L/min (IQR 60.5-44.4), VTmax 1.51L (IQR 1.88-1.21), VO2/HRmax 7.95mL/beat (IQR 9.35-6.90), and SpO2max 98% (IQR 99-97). A significant positive bivariate correlation (P<0.05) was demonstrated between CWM area, and BMI and CPET parameters.
Conclusion: Patients with SSc see a rediction in their accessory respiratory muscle volume and this correlates with a decline in cardiopulmonary function.
Limitations: The study is limited by its small sample size and single centre methodology.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information: No information provided by the submitter.

Author Block: Y. Xu¹, O. Weinheimer¹, S. Triphan¹, J. Grolig¹, O. von Stackelberg¹, J. Biederer¹, C. J. Galban², H-U. Kauczor¹, M. Wielpütz¹; ¹Heidelberg/DE, ²Ann Arbor, MI/US
Purpose: In the assessment of COPD, quantitative indices of emphysema, and functional small airways disease (fSAD), as well as perfusion were developed separately for CT and MRI, respectively. Their regional interdependencies, however, have not been assessed. The aim of this study was to utilise a machine-learning-based image registration pipeline to align pulmonary abnormalities regionally and examine voxel-wise correlations of emphysema and fSAD with quantitative perfusion MRI.
Methods or Background: Sixty-three patients from the multi-centre COSYCONET cohort underwent same-day low-dose paired inspiratory-expiratory CT as well as morphofunctional MRI. Lungs were segmented CT and on T1-weighted VIBE MRI using an in-house pipeline, separately. The segmentation was forwarded to an iterative deformable registration model, where both CT and functional 4D perfusion MR are registered to the spatial layout of VIBE MR. The correlation between quantitative CT-based parametric response maps (PRMemphysma, PRMfSAD) and MRI-based pulmonary blood flow (PBF) and perfusion defects (QDP) was studied on a voxel level. For comparing interpatient PBF, we weighted each voxel according to its relative proportion to the patient’s lung.
Results or Findings: Lung voxels labeled as PRMEmphysema, PRMfSAD, and PRMNormal in CT were classified as perfusion defects in 74.59%, 57.63%, and 30.98%, of cases respectively. Dice coefficient equals 0.681 beween PRMNormal and well perfused regions. The normalised PBF was 45.8, 56.9, and 85.6 ml/100ml/min for PRMEmphysema, PRMfSAD, and PRMNormal voxels respectively.
Conclusion: We proposed a novel registration pipeline to spatially align CT, T1-weighted MRI and 4D perfusion MR. Results suggest CT-based PRMEmphysema and PRMfSAD labels are strongly correlated to low PBF, and had a higher tendency to be classified as perfusion defects. Perfusion information is supplementary to PRM, requiring further investigation.
Limitations: The method is yet to be evaluated on a larger subgroup of the cohort.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by the Ethik-Kommission der Medizinischen Fakultät der Universität Heidelberg, S-656/2012.

Author Block: H. Ren¹, S. Dong², H. Yu¹; ¹Shanghai/CN, ²Beijing/CN
Purpose: The purpose of this study was to investigate whether there are microscopical and invisible changes in lung CT images before the detection of pulmonary metastases.
Methods or Background: Sixteen patients with pulmonary metastases were followed up for more than three months between January 2019 and June 2023. Radiomic features were extracted from high-resolution CT images before and after the visualisation of metastases. Elastic registration was applied to align images after metastasis with those before metastasis. Lung metastases were delineated using 3D-slicer, and contralateral lung tissue was outlined as contrast. Radiomic analysis was performed on the pre-metastasis images using features including first-order, grey-level co-occurrence matrix (glcm), grey-level dependence matrix (gldm), grey-level run length matrix (glrlm), grey-level size zone matrix (glszm), neighbourhood grey-tone difference matrix (ngtdm), high-order filter features, and wavelet-based features. The signatures of pre- and post-metastasis images were analysed.
Results or Findings: A total of 58 lesions met the criteria (29 metastases, 29 controls), yielding 1209 radiomic features. Wilcoxon tests identified 362 radiomic features with significant differences between pre- and post-metastasis images (P<0.05). These features were used to build a K-neighbours Classifier model. The results showed an accuracy of 0.775, sensitivity of 0.65, and specificity of 0.733 in predicting metastasis occurrence before lesion visualisation. Logistic regression analysis indicated an AUC value of 0.8.
Conclusion: Lung tissue may undergo subtle changes before the visualisation of pulmonary metastases. Radiomic features might reflect these changes earlier, but further validation is required with a larger dataset.
Limitations: Due to this study's retrospective single-centre nature, some selection bias may be unavoidable. Furthermore, all CT scans used in our study were obtained and provided by only one company.
Funding for this study: The study was funded by the National Natural Science Foundation of China (grant number 82071873), the Clinical Research Special Project of Shanghai Municipal Health Commission (grant number 20234Y0020), and the China International Medical Foundation (grant number z-2014-07-2301).
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by the Institutional Ethics Committee of Shanghai Chest Hospital, Shanghai Jiao Tong University School of Medicine. Individual consent for this retrospective analysis was waived.

Author Block: G. Kim¹, A. Swaminathan², T. Whelan², M. Neely², J. Todd², S. Palmer², D. Wojdyla², C. Conoscent³, J. Goldin¹; ¹Los Angeles, CA/US, ²Durham, NC/US, ³Ridgefield, CT/US
Purpose: Associations between quantitative measures of fibrosis on HRCT and survival in patients with progressive fibrosing ILDs are not well established. The purpose of this study was to evaluate the prognostic value of HRCT-derived scores in the ILD-PRO Registry.
Methods or Background: Patients had an ILD other than IPF, reticular abnormality and traction bronchiectasis, and met criteria for ILD progression within the prior 24 months. HRCT images taken closest to enrolment were analysed following lobar segmentation. A machine-learning algorithm derived the following quantitative scores: quantitative lung fibrosis (QLF); quantitative ground glass (QGG); quantitative honeycomb (QHC); quantitative ILD (QILD: sum of QLF, QGG and QHC). Cox proportional hazards models for time to death or lung transplant were fit and Kaplan-Meier event rates reported.
Results or Findings: Among 331 patients, median QLF, QGG, QHC and QILD scores were 14.8%, 22.7%, 0.07% and 42.3%, respectively. Event rates for death or lung transplant at 2 years in the highest vs lowest tertiles of QLF, QGG, QHC and QILD scores were 31.2% vs 19.9%, 30.4% vs 26.3%, 22.8% vs 19.4%, and 31.7% vs 17.3%, respectively. There were no significant differences in the risk of death or lung transplant across tertiles of any of the scores. The HRs (95% CI) for risk of death or lung transplant for the highest vs lowest tertiles of QLF, QGG, QHC and QILD scores were 2.20 (1.04, 4.66), 1.72 (0.88, 3.37), 0.92 (0.45, 1.90) and 2.38 (1.09, 5.20), respectively.
Conclusion: Patients with progressive fibrosing ILDs in the ILD-PRO Registry who had QLF or QILD scores in the highest tertiles showed trends towards an increased risk of death or lung transplant over 2 years compared with those with scores in the lowest tertiles.
Limitations: The study is limited by only evaluating associations between quantitative HRCT scores and short-term disease progression in patients with progressive pulmonary fibrosis.
Funding for this study: The IPF-PRO/ILD-PRO Registry is supported by Boehringer Ingelheim Pharmaceuticals, Inc (BIPI) and run in collaboration with the Duke Clinical Research Institute (DCRI) and enrolling centers.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by the Institutional Review Board of each participating centre: ClinicalTrials.gov Identifier: NCT01915511.

Author Block: J. Choi¹, K. J. Chae², H. Ko², W. Chung², G. Y. Jin², C. Lee²; ¹Kansas City, KS/US, ²Seoul/KR
Purpose: The purpose of this study was to characterise subclinical lung structure-function alteration in humidifier disinfectant-associated lung injury (HDLI) using quantitative computed tomography (qCT) and computational fluid dynamics (CFD) analysis.
Methods or Background: For 103 toxic HD-exposed patients, commercial and in-house software computed 186 multiscale lung structure-function features through quantitative inspiratory-expiratory CT image matching and one-dimensional CFD breathing simulations. We first characterised HD-associated structure-function abnormality from 72 adults with normal-appearing CTs and pulmonary function tests (PFTs), and then reviewed two-year changes in 31 patients in the national HDLI monitoring program of Korea.
Results or Findings: In HD-affected normal-appearing cases, airway narrowing resulted in a greater tracheobronchial air pressure drop particularly through the left lower lobe segmental airways (r=0.54, p<0.001). Elevated air trapping, upper and middle lobes motion, and reduced volume expansion reduced transpulmonary pressure (r=-0.69, 0.53, 0.50; p<0.001, all), while elevated high attenuation area reduced lower lobe motion (r=-0.50, p<0.001), implying no direct impact of basal lung inflammation and fibrosis on tidal breathing characteristics. Reduced whole lung motion, reduced right lower lobe ventilation, and elevated left upper lobe ventilation elevated airway resistance (r=-0.61, -0.62, 0.61; p<0.001, all). Two-year changes of these features characterised recovering, stable, and worsening subgroups of the 31 monitoring patients with no eminent changes in visual CT features and PFTs.
Conclusion: qCT-CFD analysis provided novel imaging-based regional lung structure-function characterisation of subclinical lung damage, recovery, and worsening in HDLI. This novel approach may help assess underlying mechanisms of complicated and worsening lung recovery.
Limitations: The relatively small sample size limits the applicability of the study.
Funding for this study: The study received funding from the Mid-Career Bridging Program through Seoul National University.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved with approval code: 2020-06-037, from the Jeonbuk National University Hospital.

Author Block: S. Martin, S. Mahmoudi, I. Yel, C. Booz, L. D. Grünewald, J-E. Scholtz, K. Eichler, T. Vogl, V. Koch; Frankfurt a. Main/DE
Purpose: The objective of this study was to investigate the diagnostic value of radiomics features and dual-source dual-energy CT (DECT) based material decomposition in differentiating low-risk thymomas, high-risk thymomas, and thymic carcinomas.
Methods or Background: This retrospective study included 32 patients (16 males, mean age 66±14 years) with pathologically confirmed thymic masses who underwent contrast-enhanced DECT between October 2014 and January 2023. Two experienced readers evaluated all patients regarding conventional radiomics features, as well as DECT-based texture features, including attenuation (HU), iodine density (mg/ml), and fat fraction (%). Data comparisons were performed using analysis of variance (ANOVA) and chi-square statistic tests. Receiver operating characteristic (ROC) curve analysis and Cox regression tests were used to discriminate between low-risk/ high-risk thymomas and thymic carcinomas.
Results or Findings: Of the 32 thymic tumors, 12 (38%) were low-risk thymomas, 11 (34%) were high-risk thymomas, and 9 (28%) were thymic carcinomas. Values differed significantly between low-risk thymoma, high-risk thymoma, and thymic carcinoma regarding DECT-based texture features (p≤0.023) and 30 radiomics features (p≤0.037). The area under the curve (AUC) to differentiate between low-risk/ high-risk thymomas and thymic cancer was 0.998 (95% CI, 0.915–1.000; p<0.001) for the combination of DECT imaging parameters and radiomics features, yielding a sensitivity of 100% and specificity of 96%. During a follow-up of 60 months (IQR, 35-60 months), the multiparametric approach, including radiomics features, DECT parameters, and clinical parameters, showed an excellent prognostic power to predict all-cause mortality (c-index = 0.978 (95% CI, 0.958–0.998), p=0.003).
Conclusion: A multiparametric approach including conventional radiomics features and DECT-based texture features facilitates accurate, non-invasive discrimination between low-risk/ high-risk thymomas and thymic carcinomas.
Limitations: No limitations were identified.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This retrospective study was approved by the Ethical Review Board of the publishing institution, and written informed consent was waived.