18.07.2014 Views

Segmentation of 3D Tubular Tree Structures in Medical Images ...

Segmentation of 3D Tubular Tree Structures in Medical Images ...

Segmentation of 3D Tubular Tree Structures in Medical Images ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

68 Chapter 4. Tube <strong>Segmentation</strong><br />

obta<strong>in</strong>ed from a gradient magnitude image. In this way, the voxels associated with the<br />

centerl<strong>in</strong>e <strong>of</strong> a tubular structure are segmented.<br />

Algorithm 2 Inverse gradient flow track<strong>in</strong>g tube segmentation for GVF obta<strong>in</strong>ed from<br />

gray value image.<br />

<strong>in</strong>put: GVF field V (x) with direction V n (x) and magnitude M(x)<br />

<strong>in</strong>put: centerl<strong>in</strong>e po<strong>in</strong>ts C = {x 1 , x 2 , . . . , x n }<br />

set S ← C<br />

queue Q ← C<br />

while Q ≠ {} do<br />

x ← extract(Q)<br />

for each voxel y ∈ Adj26(x) do<br />

if y /∈ S and M(y) > M(x) and argm<strong>in</strong> z∈Adj26(y) 〈V n (y), −→ yz〉 = x then<br />

S ← S ∪ {x}<br />

Q ← Q ∪ {y}<br />

end if<br />

end for<br />

end while<br />

output: segmented tube voxels S<br />

Algorithm 3 Inverse gradient flow track<strong>in</strong>g tube segmentation for GVF obta<strong>in</strong>ed from<br />

gradient magnitude image.<br />

<strong>in</strong>put: GVF field V (x) with direction V n (x) and magnitude M(x)<br />

<strong>in</strong>put: centerl<strong>in</strong>e po<strong>in</strong>ts C = {x 1 , x 2 , . . . , x n }<br />

set S ← C<br />

queue Q ← C<br />

while Q ≠ {} do<br />

x ← extract(Q)<br />

for each voxel y ∈ Adj26(x) do<br />

if y /∈ S and argm<strong>in</strong> z∈Adj26(y) 〈V n (y), −→ yz〉 = x then<br />

S ← S ∪ {x}<br />

Q ← Q ∪ {y}<br />

end if<br />

end for<br />

end while<br />

output: segmented tube voxels S<br />

4.3 Graph Cut Based <strong>Tree</strong> <strong>Segmentation</strong><br />

In this section, we present an approach to accurately del<strong>in</strong>eate the boundary <strong>of</strong> tree structures<br />

<strong>in</strong> a globally optimal manner us<strong>in</strong>g a graph cut segmentation approach [17]. The

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!