© RSNA, 2012
The Qanadli (6) score was calculated by assuming 10 segmental pulmonary arteries and assigning a score of 1 to an embolus in each segmental pulmonary artery. An embolus in a more proximal artery was given the value of all the segmental arteries distal to the affected pulmonary artery. A weighting factor was then applied that was related to the amount of occlusion (0, no defect; 1, partial occlusion; 2, complete occlusion). A subsegmental embolus was recorded as a partially occluded segmental pulmonary artery. The maximum possible score was 40 (6). The Mastora score was calculated on the basis of five mediastinal pulmonary arteries, six lobar pulmonary arteries, and 20 segmental pulmonary arteries. A severity score was applied to each clot based on the percentage of occlusion (score 1, < 25%; score 2, 25%-49%; score 3, 50%-74%; score 4, 75%-99%; and score 5, 100%). The maximum possible score was 155 (7).
We developed a semiautomated program for the segmentation and quantification of pulmonary embolus volume from CT pulmonary angiographic images. The program was developed on the basis of a modified seeded region-growing algorithm and implemented on a personal computer running a Windows operating system (Microsoft, Redmond, Wash). The semiautomated segmentation method involves two steps: (a) placement of a “seed” over an embolus identified by an observer and (b) propagation of the seed to neighboring voxels with Hounsfield units similar to those of the seed and automated segmentation by the computer. While the first step relies on the identification of a clot by the observer, the second step takes advantage of the reliability of the computer (19).