Limits...
Segmentation of Pulmonary Vascular Trees from Thoracic 3D CT Images.

Shikata H, McLennan G, Hoffman EA, Sonka M - Int J Biomed Imaging (2009)

Bottom Line: A quantitative validation was performed with more than 1000 manually identified points selected from inside the vessel segments to assess true positives (TPs) and 1000 points randomly placed outside of the vessels to evaluate false positives (FPs) in each case.On average, for both the high and low volume lung images, 99% of the points was properly marked as vessel and 1% of the points were assessed as FPs.Our hybrid segmentation algorithm provides a highly reliable method of segmenting the combined pulmonary venous and arterial trees which in turn will serve as a critical starting point for further quantitative analysis tasks and aid in our overall goal of establishing a normative atlas of the human lung.

View Article: PubMed Central - PubMed

Affiliation: Iowa Institute for Biomedical Imaging, The University of Iowa, Iowa City, IA 52242, USA.

ABSTRACT
This paper describes an algorithm for extracting pulmonary vascular trees (arteries plus veins) from three-dimensional (3D) thoracic computed tomographic (CT) images. The algorithm integrates tube enhancement filter and traversal approaches which are based on eigenvalues and eigenvectors of a Hessian matrix to extract thin peripheral segments as well as thick vessels close to the lung hilum. The resultant algorithm was applied to a simulation data set and 44 scans from 22 human subjects imaged via multidetector-row CT (MDCT) during breath holds at 85% and 20% of their vital capacity. A quantitative validation was performed with more than 1000 manually identified points selected from inside the vessel segments to assess true positives (TPs) and 1000 points randomly placed outside of the vessels to evaluate false positives (FPs) in each case. On average, for both the high and low volume lung images, 99% of the points was properly marked as vessel and 1% of the points were assessed as FPs. Our hybrid segmentation algorithm provides a highly reliable method of segmenting the combined pulmonary venous and arterial trees which in turn will serve as a critical starting point for further quantitative analysis tasks and aid in our overall goal of establishing a normative atlas of the human lung.

No MeSH data available.


Related in: MedlinePlus

Average number of false (extra) branches and missing branches of the phantom evaluated by graph representation of the segmentation results. The false branches and missing branches were counted manually. (a) A total number of false branches (false positives). (b) A total number of missing branches (true negatives).
© Copyright Policy - open-access
Related In: Results  -  Collection


getmorefigures.php?uid=PMC2801012&req=5

fig7: Average number of false (extra) branches and missing branches of the phantom evaluated by graph representation of the segmentation results. The false branches and missing branches were counted manually. (a) A total number of false branches (false positives). (b) A total number of missing branches (true negatives).

Mentions: The segmentation algorithm has one major parameter, the threshold of the filter output used to obtain the initial segmentation. When it is decreased, more vessels and noise elements are extracted, increasing the TP rate as well as the FP rate. On the contrary, less vessels (and noise elements) are extracted if the threshold is increased, causing a lowering of the TP and FP rates. Since the threshold is a tradeoff between TP and FP rates, four thresholds were selected empirically and were tested by use of the simulation. Figure 7 shows average number of false (extra) branches and missing branches as a function of the threshold value. When the threshold setting was 0.06, the segmentation results were missing less than one branch per volume on average. However, the number of extra branches increased rapidly when noise levels went up. On the contrary, when the threshold was 0.09, the results missed 3.5 to 4.5 out of 125 branches on average whereas they had less extra branches than other threshold settings. In [8], a noise level σ = 20 was used to represent typical noise in CT images. For comparison, σ = 40 represents noise in ultrasound images. When the threshold was 0.07, the algorithm missed less than one branch on average for all noise levels and the results contained one or less extra branches in the case of σ = 20, 30. Derived from this result, a threshold value of 0.07 was used for the following experiments using the clinical data sets.


Segmentation of Pulmonary Vascular Trees from Thoracic 3D CT Images.

Shikata H, McLennan G, Hoffman EA, Sonka M - Int J Biomed Imaging (2009)

Average number of false (extra) branches and missing branches of the phantom evaluated by graph representation of the segmentation results. The false branches and missing branches were counted manually. (a) A total number of false branches (false positives). (b) A total number of missing branches (true negatives).
© Copyright Policy - open-access
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC2801012&req=5

fig7: Average number of false (extra) branches and missing branches of the phantom evaluated by graph representation of the segmentation results. The false branches and missing branches were counted manually. (a) A total number of false branches (false positives). (b) A total number of missing branches (true negatives).
Mentions: The segmentation algorithm has one major parameter, the threshold of the filter output used to obtain the initial segmentation. When it is decreased, more vessels and noise elements are extracted, increasing the TP rate as well as the FP rate. On the contrary, less vessels (and noise elements) are extracted if the threshold is increased, causing a lowering of the TP and FP rates. Since the threshold is a tradeoff between TP and FP rates, four thresholds were selected empirically and were tested by use of the simulation. Figure 7 shows average number of false (extra) branches and missing branches as a function of the threshold value. When the threshold setting was 0.06, the segmentation results were missing less than one branch per volume on average. However, the number of extra branches increased rapidly when noise levels went up. On the contrary, when the threshold was 0.09, the results missed 3.5 to 4.5 out of 125 branches on average whereas they had less extra branches than other threshold settings. In [8], a noise level σ = 20 was used to represent typical noise in CT images. For comparison, σ = 40 represents noise in ultrasound images. When the threshold was 0.07, the algorithm missed less than one branch on average for all noise levels and the results contained one or less extra branches in the case of σ = 20, 30. Derived from this result, a threshold value of 0.07 was used for the following experiments using the clinical data sets.

Bottom Line: A quantitative validation was performed with more than 1000 manually identified points selected from inside the vessel segments to assess true positives (TPs) and 1000 points randomly placed outside of the vessels to evaluate false positives (FPs) in each case.On average, for both the high and low volume lung images, 99% of the points was properly marked as vessel and 1% of the points were assessed as FPs.Our hybrid segmentation algorithm provides a highly reliable method of segmenting the combined pulmonary venous and arterial trees which in turn will serve as a critical starting point for further quantitative analysis tasks and aid in our overall goal of establishing a normative atlas of the human lung.

View Article: PubMed Central - PubMed

Affiliation: Iowa Institute for Biomedical Imaging, The University of Iowa, Iowa City, IA 52242, USA.

ABSTRACT
This paper describes an algorithm for extracting pulmonary vascular trees (arteries plus veins) from three-dimensional (3D) thoracic computed tomographic (CT) images. The algorithm integrates tube enhancement filter and traversal approaches which are based on eigenvalues and eigenvectors of a Hessian matrix to extract thin peripheral segments as well as thick vessels close to the lung hilum. The resultant algorithm was applied to a simulation data set and 44 scans from 22 human subjects imaged via multidetector-row CT (MDCT) during breath holds at 85% and 20% of their vital capacity. A quantitative validation was performed with more than 1000 manually identified points selected from inside the vessel segments to assess true positives (TPs) and 1000 points randomly placed outside of the vessels to evaluate false positives (FPs) in each case. On average, for both the high and low volume lung images, 99% of the points was properly marked as vessel and 1% of the points were assessed as FPs. Our hybrid segmentation algorithm provides a highly reliable method of segmenting the combined pulmonary venous and arterial trees which in turn will serve as a critical starting point for further quantitative analysis tasks and aid in our overall goal of establishing a normative atlas of the human lung.

No MeSH data available.


Related in: MedlinePlus