Quantification of regional deformation of the lungs by non-rigid registration of three-dimensional contrast-enhanced magnetic resonance imaging.

BACKGROUND

Assessment of lung function is vital for the diagnosis of a variety of pathological conditions. Research has been proposed to study pulmonary mechanics and kinematics using two-dimensional (2D) magnetic resonance imaging (MRI). This allows estimation of regional lung tissue mechanics but is limited to 2D information. An approach based on three-dimensional (3D) contrast-enhanced MR angiogram of pulmonary blood vessels and a non-rigid image registration technique is proposed for quantification of lung regional deformations, which can potentially be used for assessment of pulmonary parenchymal mechanics and regional ventilation for disease diagnosis without ionizing radiation.

METHODS

On three volunteers, an end-expiration scan and end-inspiration scan was acquired successively for each volunteer using a 3D breath-hold contrast-enhanced MRI sequence several minutes after gadolinium injection. Subsequently, a rectangle box lung mask is manually selected for each end-expiration scan, applying non-rigid registration algorithms using cubic B-splines as transformations to align each pair of images. This incorporates the Normalized Correlation Coefficient similarity with the bending energy term as cost function with a multi-resolution multi-grid approach. Finally, the lung regional 3D deformations were obtained using the transformations obtained by registration. The alignment accuracy after non-rigid registration was estimated by using a set of branch points of pulmonary blood vessels as anatomical landmarks for each pair of images.

RESULTS

With contrast enhancement, the pulmonary blood vessel signal was enhanced, which greatly facilitated the non-rigid registration in the lung parenchyma. The average landmarks distances in three pairs of datasets are reduced from 17.9, 20.3 and 16.3 mm, to 1.0, 1.6 and 1.2 mm, respectively, by non-rigid registration. After registration, the average distances error of each pair of datasets was less than 0.6 mm in the right-to-left (RL) direction, less than 0.9 mm in the inferior-to-superior (IS) direction, and less than 1.2 mm in the anterior-to-posterior (AP) direction.

CONCLUSIONS

Results demonstrated that the proposed method can accurately register lungs with large deformations to evaluate lung regional deformation. It may be used for quantitative assessment of 3D lung regional ventilation avoiding ionizing radiation.