Lung function measurement of single lungs by lung area segmentation on 2D dynamic MRI

Rationale and objectives: Most lung disease is inhomogeneously distributed but diagnosed by global spirometry. Regional lung function might allow for earlier diagnosis. Dynamic two-dimensional magnetic resonance imaging (2D-MRI) can depict lung motion with high temporal resolution. We evaluated whether measurement of lung area on dynamic 2D-MRI has sufficient agreement with spirometry to allow for lung function testing of single lungs.

Material and methods: Ten healthy volunteers were examined in a 1.5 T MRI scanner with a Flash 2D-sequence (8.5 images per second, sagittal and coronal orientation) with simultaneous spirometry. The lung area was segmented semiautomatically and the area changes were compared with spirometric volume changes.

Results: Segmentation of one time series took 191 seconds on average. Volume-time and flow-volume curves from MRI data were almost congruent with spirometric curves. Pearson correlation of MRI area with spirometry was very high (mean correlation coefficients >0.97). Bland-Altman plots showed good agreement of lung area with spirometry (95% limits of agreement below 11% in each direction). Differences between lung area and spirometry were significantly smaller for sagittal measurement of the right lung than sagittal measurement of the left lung and coronal measurement. The relative forced expiratory volume in the first second differed less than 5% between MRI and spirometry in all but one volunteer.

Conclusions: Measurement of lung area on 2D-MRI allows for functional measurement of single lungs with good agreement to spirometry. Postprocessing is fast enough for application in a clinical context and possibly provides increased sensitivity for lung functional measurement of inhomogeneously distributed lung disease.