Modeling Respiratory Signals by Deformable Image Registration on 4DCT Lung Images

Algorithm 2

Candidate lung partition segmentation.

1. Input the SliceIM (One lung slice image in a phase without artifacts)
2. Within Class Variance approach to binarize an image
BinaryIM = WithinClassVariance(SliceIM)
3. Complement of the BinaryIM to change the area of interest into 1 and background into
0 with BinaryIM =1 – BinaryIM;
4. Fill all holes in BinaryIM and keep these large areas (in this case the large area is greater than 30)
BinaryIM = FillHoles(BinaryIM)
BinaryIM = LargeArea(BinaryIM, 30)
(1)	5. Store the result as a Candidate Lung partition
CandidateIM = BinaryIM
6. End
7. Result in CandidateIM (The candidate lung partitions)
Appendix
WithinClassVariance method
1. Compute histograms and probabilities of each intensity level
2. Set up initial ωi(0) and μi(0)
3. Step through all possible thresholds t =1, .. maximum intensity
a. Update ωi(0) and μi(0)
b. Compute σ^2b(t)
4. Desired threshold corresponds to the maximum σ^2b(t)