|
Pulmonary embolism (PE) is a medical condition defined as the obstruction of pulmonary arteries by a blood
clot, usually originating in the deep veins of the lower limbs. PE is a common but elusive illness that can cause
significant disability and death if not promptly diagnosed and effectively treated. CT Pulmonary Angiography
(CTPA) is the first line imaging study for the diagnosis of PE. While clinical prediction rules have been recently
developed to associate short-term risks and stratify patients with acute PE, there is a dearth of objective biomarkers
associated with the long-term prognosis of the disease. Clot (embolus) burden is a promising biomarker for the
prognosis and recurrence of PE and can be quantified from CTPA images. However, to our knowledge, no study
has reported a method for segmentation and measurement of clot from CTPA images. Thus, the purpose of this
study was to develop a semi-automated method for segmentation and measurement of clot from CTPA images. Our
method was based on Modified Seeded Region Growing (MSRG) algorithm which consisted of two steps: (1) the
observer identifies a clot of interest on CTPA images and places a spherical seed over the clot; and (2) a region
grows around the seed on the basis of a rolling-ball process that clusters the neighboring voxels whose CT
attenuation values are within the range of the mean ± two standard deviations of the initial seed voxels. The rollingball
propagates iteratively until the clot is completely clustered and segmented. Our experimental results revealed
that the performance of the MSRG was superior to that of the conventional SRG for segmenting clots, as evidenced
by reduced degrees of over- or under-segmentation from adjacent anatomical structures. To assess the clinical value
of clot burden for the prognosis of PE, we are currently applying the MSRG for the segmentation and volume
measurement of clots from CTPA images that are acquired in a large cohort of patients with PE in an on-going
NIH-sponsored clinical trial.
|
