14.08.2013 Views

Docteur de l'université Automatic Segmentation and Shape Analysis ...

Docteur de l'université Automatic Segmentation and Shape Analysis ...

Docteur de l'université Automatic Segmentation and Shape Analysis ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Chapter 3 Hippocampal segmentation using multiple atlases 57<br />

the similarity measurement between two atlases<br />

SimNMI(Ai, Aj) = 1 (<br />

2<br />

SimC(Ai, Aj) = 1<br />

2<br />

3.3.1.2 Atlas selection by MMR algorithm<br />

Sim A<br />

NMI(Ai, Ij) + Sim A<br />

NMI(Aj, Ii) )<br />

(3.22)<br />

(<br />

Sim A<br />

C(Ai, Ij) + Sim A<br />

C(Aj, Ii) )<br />

. (3.23)<br />

We initialize the set of selected atlases A to be empty <strong>and</strong> select one atlas each<br />

iteration. At each iteration, the atlas ˆ k is selected according to MMR, such that<br />

(<br />

ˆk = arg max λSim<br />

k /∈A<br />

A<br />

(·)(Ak, I) − (1 − λ) max<br />

j∈A Sim(·)(Aj,<br />

)<br />

Ak) , (3.24)<br />

until the selected atlas set A reaches a preset threshold. The similarity Sim A<br />

(·) <strong>and</strong><br />

Sim(·) are to be substituted in practice by the metrics <strong>de</strong>scribed previously. The<br />

parameter λ ∈ (0, 1] controls the similarity measurement Sim A<br />

(·)(·, ·). In addition<br />

to similarity, diversity is introduced by penalizing the redundancy<br />

max<br />

j∈A Sim(·)(Aj, Ak) (3.25)<br />

within the selected set A . When λ = 1, MMR is equivalent to the similarity<br />

ranking using Sim A<br />

(·)(·, ·).<br />

The algorithm of atlas selection using MMR is listed in Algorithm 3. The matrix<br />

of atlas similarity (<br />

Sim(·)(Ai, Aj) )<br />

can be pre-computed <strong>and</strong> stored.<br />

i,j=1,··· ,n<br />

Algorithm 3 Atlas re-ranking by MMR<br />

1: ˆ k ← arg max Sim<br />

k<br />

A<br />

(·)(Ak, I)<br />

2: A ← { ˆ k}<br />

3: while |A | ≤ the ( number of atlases to be selected do<br />

4: kˆ ← arg max λSim<br />

k /∈A<br />

A<br />

5: A ← A ∪ { ˆ k}<br />

6: end while<br />

(·)(Ak, I) − (1 − λ) maxj∈A Sim(·)(Aj, Ak) )

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

Saved successfully!

Ooh no, something went wrong!