Aims and objectives
This paper introduces a framework for automatic classification of pulmonary artery and vein sub-trees.
Despite the findings that there is a relation between vessel distribution and various pulmonary diseases,
there has not been much research about quantification and classification of pulmonary artery and vein sub-trees in vivo, due to their morphological difficulties in volumetric CT.
Therefore,
the aim of this study isto make and evaluate automatic classifier by tree modeling and extract quantitative features from the tree structure in viewpoints of geometry and topology from...
Methods and materials
Non-contrast volumetric chest CT scans with sub-millimeter thickness of 10 normal subjects were used.
We extracted pulmonary vessels using a thresholding method (-700 HU) from CT images.
The hilar region was removed from the initial vessel to divide the vessel tree into separated sub-trees.
The skeletons of the tree having mixed artery and vein were extracted using a 3D thinning method,
and minimum spanning tree algorithm were used to differentiate the artery and vein tree separately.
In addition,
the radius estimation on each skeleton voxel...
Results
235 arteries and 191 veins from 10 normal subjects were included in this study.
Uni-variate analysis revealed that mean and SD of diameter,
branch length,
and tapering ratio,
and mean branch asymmetry among the features shows significant different (t-test,
all p<0.04).
The overall classification accuracy with cross validation was 80.37±1.46%.
Conclusion
Based on geometric and topologic features of artery and vein tree,
the classifier could automatically differentiate artery and vein sub-trees with reasonable accuracy.This method could be used for automatic artery and vein color-coding for chest radiologist,
which could be helpful for lessening radiologists’ workload.
References
[1] S.
Park,
S.
M.
Lee,
N.
Kim et al.,
“Automatic reconstruction of the arterial and venous trees on volumetric chest CT,” Med Phys, vol.
40,
no.
7,
pp.
071906,
2013.