This example segments a 3D Computed Topography (CT) image of the spine. Therefore the
SimITKLibrarySS3D is used.
- Open the SimITKLibrarySS3D.
- Create a new model.
- Drag the itkReaderSS3D into the model window. Double click on it and input the name of
the file that will be read in.
- The background of the spine volumes is very grainy and noisy. This will interfere
in the segmenting, therefore the images have to be blurred before thresholding. Drag the
itkBinomialBlurImageFIlterSS3D block into the model window. The binomial blur image filter
computes a nearest neighbor average along each dimension. This process is repeated as
many times as you want.
- Double click on the itkBinomialBlurImageFIlterSS3D block and input the number of repetitions.
In this example 4 repetitions were performed.
- Connect the input block to the binomial blur block.
- Drag the itkWriterSS3D to the model window.
- Connect the binomial blur block to the writer.
- Double click on the writer and input the name of the output file. Example: ‘Binomial.mhd’
- Run the model by clicking on the run button on the toolbar.
- Open the output file with OCCI viewer to see the effect of the binomial blur on the CT volumes.
- Drag the itkBinaryThresholdImageFilterSS3D into the model window. Double click the
threshold block. The filter sets the image intensity to the inside value if it is between
the lower and the upper threshold. The intensity is set to the outside value otherwise.
- Input the parameters:
- Inside value = 1
- Lower threshold = 100
- Outside value = 0
- Upper threshold = 255
- Delete the arrows between the binomial blur block and the writer. Click on the arrow and press delete.
- Connect the itkBinomialBlurImageFilterSS3D block to the itkBinaryThresholdImageFilterSS3D block.
- Connect the threshold block to the writer.
- Double click on the writer and input the name of the output file. Example: ‘spineThreshold.vtk’.
- Open the file in OCCI viewer.
Segmenting Using Binomial Blur and Region Growing
This example segments a 3D Computed Topography (CT) image of the spine therefore the
SimITKLibrarySS3D is used. Instead of thresholding, region growing is used to achieve a
full-coloured vertebrae rather than just an outline. As the example above, the image is
first blurred with a binomial blur filter, using 4 repetitions.
- Create a model that has an itkReaderSS3D, itkBinomialBlurImageFilterSS3D,
itkConfidenceConnectgedImageFilterSS3D and an itkWriterSS3D. Connect the blocks
respectively (click to enlarge image).
- Double-click the itkReaderSS3D and input the name of the input file. Example: ‘phantomCT.vtk’
- Double-click the itkBinomialBlurImageFilterSS3D and input 4 as the number of repitions.
- Double-click the itkConfidenceConnectedImageFilterSS3D and input the parameters:
- Initial Neighborhood Radius: 3
- Multiplier: 4.5
- Number of iterations: 4
- Replace value: 225
- Seed = [17.41 26.72 41.60]
- Double-click on the writer and input the name of the output file. Example: ‘SpineConnected.mhd’.
- Open the output file using OCCI viewer.
> Continue onto the Surface Model tutorial.