Preview

PDF, English Download (4MB) | Terms of use

Abstract

Pulmonary perfusion is regionally impaired in muco-obstructive lung diseases such as cystic fibrosis (CF) and chronic obstructive pulmonary disease (COPD) due to the destruction of the alveolar-capillary bed and hypoxic pulmonary vasoconstriction in response to alveolar hypoxia. Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is an established technique for assessing regional perfusion abnormalities by exploiting contrast enhancement in the lung parenchyma during the first pass of an intravenously injected contrast agent bolus. Typically, perfusion abnormalities are assessed in clinical studies by visual scoring or by quantifying pulmonary blood flow (PBF) and pulmonary blood volume (PBV). Automated quantification can help to address inter-reader variability issues with human reader, facilitate detailed perfusion analyses and is time efficient. However, currently used absolute quantification of PBF and PBV is highly variable. For this reason, an algorithm was developed to quantify the extent of pulmonary perfusion in percent (QDP) using unsupervised clustering algorithms, which leads to an intrinsic normalisation and can reduce variability compared to absolute perfusion quantification. The aims of this work were to develop a robust algorithm for quantifying QDP, to investigate the midterm reproducibility of QDP, and to validate QDP using MRI perfusion scoring, quantitative computed tomography (CT) parameters, and pulmonary function testing (PFT) parameters. Furthermore, the performance of QDP was compared to the performance of PBF and PBV. The development of QDP and its technical and clinical validation were performed using data from two studies, which utilise DCE-MRI. First, the algorithm was developed using data of 83 COPD subjects from the ‘COSYCONET’ COPD cohort by comparing different unsupervised clustering approaches including Otsu´s method, k-means clustering, and 80th percentile threshold. Second, the reproducibility of QDP was investigated using data from a study of 15 CF and 20 COPD patients who underwent DCE-MRI at baseline and one month later (reproducibility study). According to the indicator dilution theory, impulse response function maps were calculated from DCE-MRI data, which formed the basis for the quantification of QDP, PBF and PBV. Overall, QDP based on Otsu´s method showed the highest agreement with the MRI perfusion score, quantitative CT parameters and PFT parameters in the COSYCONET study and was therefore selected for further evaluations. QDP correlated moderately with the MRI perfusion score in CF (r=0.46, p<0.05) and moderately to strongly in COPD (r=0.66 and r=0.72, p<0.001) in both studies. PBF and PBV correlated poorly with the MRI perfusion score in CF (r=-0.29, p=0.132 and r=-0.35, p=0.067, respectively) and moderately in COPD (r=-0.49 to -0.57, p<0.001). QDP correlated strongly with the CT parameter for emphysema (r=0.74, p<0.001) and weakly with the CT parameter for functional small airway disease (r=0.35, p<0.001) in COPD. The extent of perfusion defects from DCE-MRI corresponded to extent of abnormal lung (emphysema+functional small airway disease) from CT, with a mean difference of 6.03±16.94. QDP correlated moderately with PFT parameters in both studies and patient groups, with one exception in the reproducibility study where no correlation was observed in the COPD group. The use of unsupervised clustering approaches increased the reproducibility (±1.96SD related to the median) of QDP (CF: ±38%, COPD: ±37%) compared to PBF(CF: ±89%, COPD: ±55%) and PBV(CF: ±55%, COPD: ±51%) and reduced outliers. These results demonstrate that the quantification of pulmonary perfusion using unsupervised clustering approaches in combination with the mathematical models of the indicator dilution theory improves the reproducibility and the correlations with visual MRI perfusion scoring, quantitative CT parameters and PFT parameters. QDP based on Otsu´s method showed high agreement with the MRI perfusion score, suggesting that in future clinical studies pulmonary perfusion can be assessed objectively by computer algorithms replacing the time-consuming visual scoring. Concordance between the extent of QDP from MRI and the extent of abnormal lung from CT indicates that pulmonary perfusion abnormalities themselves may contribute to, or at least precede, the development of irreversible emphysema. The findings of both studies show that QDP is clinically meaningful in muco-obstructive lung diseases as it is significantly associated with the MRI perfusion score, quantitative CT parameters, and PFT parameters.