Introduction to the Insight Toolkit (ITK)¶
Learning Objectives¶
- Learn how to run cells in a Jupyter Notebook
- Run a segmentation example that demonstrates ITK's ability to provide insight into images
- Understand the purpose and capabilities of the toolkit
Jupyter Notebooks¶
These are Jupyter Notebooks, an open-source web application that allows you to create and share documents that contain live code, equations, visualizations and narrative text.
To run cells in the notebook, press shift + enter.
For more information, see the Notebook Help
Insight Into Images¶
In [1]:
import itk
from itkwidgets import view
In [2]:
file_name = 'data/brainweb165a10f17.mha'
image = itk.imread(file_name, itk.ctype('float'))
view(image, slicing_planes=True, gradient_opacity=0.8, ui_collapsed=True)
Viewer(gradient_opacity=0.8, rendered_image=<itkImagePython.itkImageF3; proxy of <Swig Object of type 'itkImag…
In [3]:
# Smooth the image
smoothed = itk.curvature_flow_image_filter(image,
number_of_iterations=6,
time_step=0.005)
view(smoothed, slicing_planes=True, gradient_opacity=0.8, ui_collapsed=True)
Viewer(gradient_opacity=0.8, rendered_image=<itkImagePython.itkImageF3; proxy of <Swig Object of type 'itkImag…
In [4]:
# Segment the white matter with a 3D region-growing algorithm
confidence_connected = itk.ConfidenceConnectedImageFilter.New(smoothed)
confidence_connected.SetMultiplier(2.5)
confidence_connected.SetNumberOfIterations(5)
confidence_connected.SetInitialNeighborhoodRadius(2)
confidence_connected.SetReplaceValue(255)
confidence_connected.AddSeed([118, 133, 92])
confidence_connected.AddSeed([63, 135, 94])
confidence_connected.AddSeed([63, 157, 90])
confidence_connected.AddSeed([111, 150, 90])
confidence_connected.AddSeed([111, 50, 88])
In [5]:
confidence_connected.Update()
view(confidence_connected, ui_collapsed=True, cmap='BuPu', shadow=False, annotations=False)
Viewer(annotations=False, cmap='BuPu', gradient_opacity=0.22, rendered_image=<itkImagePython.itkImageSS3; prox…
History of ITK¶
History¶
In 1999, the US National Institute of Health’s (NIH) National Library of Medicine (NLM) started a project to support the Visible Human Project.
Goals¶
- Collect best-of-the-best image analysis algorithms for reproducible science.
- Provide a software resource for teaching medical image analysis algorithms.
- Establish a foundation for future research.
- Support commercial applications.
- Create conventions for future work.
- Grow a self-sustaining community of software users and developers.
Continued Development¶
- Development has progressed since 1999
- Contributions from over 267 developers
- Over 1.6 million lines of code
- Downloaded over 500,000 times
ITK contributors locations for the 4.8 and 4.9 releases.
Features¶
N-Dimensional Image Filtering¶
Filtering Algorithms Classes¶
- Fast marching methods
- Convolution
- Image gradient
- Denoising
- Thresholding
- Mathematical morphology
- Smoothing
- Image features
- Image statistics
- Bias correction
- Image grid operations
- ....
Segmentation¶
Segmentation¶
Segmentation¶
Segmentation¶
Segmentation¶
Segmentation¶
Registration¶
ITK Resources¶
ITK Software Guide: https://www.itk.org/ItkSoftwareGuide.pdf
Discourse Discussion: https://discourse.itk.org
Sphinx Examples: https://www.itk.org/ITKExamples
Examples/directory in the ITK repositoryWiki Examples: http://www.itk.org/Wiki/ITK/Examples
Kitware: https://www.kitware.com/
Kitware, Inc¶
- Collaborative, Open Source, Scientific Software Development
- Carrboro, NC, near University of North Carolina-Chapel Hill, in the Research Triangle