Elastic Registration Algorithm Based on Three-dimensional Pulmonary MRI in Quantitative Assessment of Severity of Idiopathic Pulmonary Fibrosis

Abstract

Purpose: To quantitatively analyze lung elasticity in idiopathic pulmonary fibrosis (IPF) using elastic registration based on 3-dimensional pulmonary magnetic resonance imaging (3D-PMRI) and to assess its’ correlations with the severity of IPF patients. Material and Methods: Thirty male patients with IPF (mean age: 62±6 y) and 30 age-matched male healthy controls (mean age: 62±6 y) were prospectively enrolled. 3D-PMRI was acquired with a 3-dimensional ultrashort echo time sequence in end-inspiration and end-expiration. MR images were registered from end-inspiration to end-expiration with the elastic registration algorithm. Jacobian determinants were calculated from deformation fields on color maps. The log means of the Jacobian determinants (Jac-mean) and Dice similarity coefficient were used to describe lung elasticity between 2 groups. Then, the correlation of lung elasticity with dyspnea Medical Research Council (MRC) score, exercise tolerance, health-related quality of life, lung function, and the extent of pulmonary fibrosis on chest computed tomography were analyzed. Results: The Jac-mean of IPF patients (−0.19, [IQR: −0.22, −0.15]) decreased (absolute value), compared with healthy controls (−0.28, [IQR: −0.31, −0.24], P<0.001). The lung elasticity in IPF patients with dyspnea MRC≥3 (Jac-mean: −0.15; Dice: 0.06) was significantly lower than MRC 1 (Jac-mean: −0.22, P=0.001; Dice: 0.10, P=0.001) and MRC 2 (Jac-mean: −0.21, P=0.007; Dice: 0.09, P<0.001). In addition, the Jac-mean negatively correlated with forced vital capacity % (r=−0.487, P<0.001), forced expiratory volume 1% (r=−0.413, P=0.004), TLC% (r=−0.488, P<0.001), diffusing capacity of the lungs for carbon monoxide % predicted (r=−0.555, P<0.001), 6-minute walk distance (r=−0.441, P=0.030) and positively correlated with respiratory symptoms (r=0.430, P=0.042). Meanwhile, the Dice similarity coefficient positively correlated with forced vital capacity % (r=0.577, P=0.004), forced expiratory volume 1% (r=0.526, P=0.012), diffusing capacity of the lungs for carbon monoxide % predicted (r=0.435, P=0.048), 6-minute walk distance (r=0.473, P=0.016), final peripheral oxygen saturation (r=0.534, P=0.004), the extent of fibrosis on chest computed tomography (r=−0.421, P=0.021) and negatively correlated with activity (r=−0.431, P=0.048). Conclusion: Lung elasticity decreased in IPF patients and correlated with dyspnea, exercise tolerance, health-related quality of life, lung function, and the extent of pulmonary fibrosis. The lung elasticity based on elastic registration of 3D-PMRI may be a new nonradiation imaging biomarker for quantitative evaluation of the severity of IPF.