Robust lung nodule growth measurement by combining registration and segmentation
S. Kabus and F. Müller and R. Wiemker and B. Fischer
    15--23  (2008)

Growth assessment for lung nodules is known to be the most relevant clinical parameter to distinguish malignant nodules from benign ones. The assessment of growth is usually done by multiple segmentations in in follow-up CT examinations of a patient. Each segmentation is performed separately yielding a volume number assigned to the nodule at a certain point in time. Experiments have shown that the segmented volume may change due to small variations in certain voxel values. To circumvent this ill-posed problem, we present a growth assessment scheme which combines the result of an elastic registration and of a segmentation method. Regarding the absence of publicly available ground truth data, a method for the creation of artificial nodules is developed. Experiments show that the described combined growth assessment method leads to similar or better results than a growth assessment method achieved by a segmentation method only.