Matching lung volume data sets--a novel approach.

There is a significant demand in matching CT datasets of the lung. The increasing number of CT slices per examination due to the higher resolution of modern CT scanners and the need for quantification of the progress of disease and healing processes in follow-up studies. A volunteer's lung was scanned by the means of multidetector CT in two different states of ventilation. The necessary lung structures for the matching procedure like lung surface and branching points were segmented. A thin-plate spline method was used to calculate the matched lung volume. The preliminary results show an average error of 2 voxel, i.e. 2mm. The calculation of the transformation matrix takes about one second on a conventional PC, which is considerably faster than other methods described in literature. The method described may be apt to be introduced in radiological practice when it comes to compare high resolution CT scans in follow-up studies quantitatively.