Frequently asked questions (FAQ)

General information and availability

Do I need VTK to build ITK?

With the default configuration, ITK does not require VTK. However, some optional modules (e.g. ITKVTKGlue) do require VTK.

Data formats

What 3D data file formats can ITK import and export?

The following table identifies the file formats that ITK can read/write via a factory mechanism based on the file extension (itk::ImageIOFactory). Some proprietary files format are only imported.

More information on supported file formats is available on the ITK-Wasm Image File Formats site.

Other file formats not listed below, such as OME-NGFF, may be supported by ITK’s remote modules. See for example the ITKIOOMEZarrNGFF remote module for Zarr-backed OME-NGFF read/write capabilities.

itkwidgets offers a Python interface for interactive visualization on the web of all of these images.

† BMP, PNG and JPEG are not very good formats for 3D medical images. They only support 2D images and a limited set of pixel types such as unsigned char, and unsigned short. The great advantage of BMP, PNG and JPEG is that you can view them with almost any image viewer. It is possible to read/write 3D datasets as collections of 2D slices each one in an independent BMP, PNG or JPEG file by using the ImageSeriesReader and itk::ImageSeriesWriter.

‡ It is higly recommended not using this format. RAW is not a format, it is insufficient, inconsistent and simply dangerous. A RAW file without a header is useless, and once you add a header, it is not RAW anymore.

What if my file format is not supported by ITK?

If ITK doesn’t have a specific file format reader at this point, you may try converting this specific file format image to a format like PNG, for which ITK has readers.

The imageio Python package contains a good set of tools for performing such conversions.

You can also take advantage of the VTK readers/importers. All you need is then use the VTKImageToImage adaptor in order to convert the vtkImageData into an itk::Image.

Supported VTK file formats can be found in https://docs.vtk.org/en/latest/supported_data_formats.html.

What mesh and point set file formats can ITK import and export?

ITK supports reading and writing the following mesh and point set file formats:

More information on supported mesh and point set file formats is available on the ITK-Wasm Mesh and Point Set File Formats site.

What transform file formats can ITK import and export?

ITK supports reading and writing the following transform file formats:

File Format

Read/Write

Import

InsightLegacy

itk::TxtTransformIOTemplate

ITK HDF5

itk::HDF5TransformIOTemplate

MATLAB

itk::MatlabTransformIOTemplate

MINC (Medical Image NetCDF)

itk::MINCTransformIOTemplate

DICOM data

How do I read a volume from a DICOM series?

The following are alternative options for reading DICOM series. The first one in the list is the recommended option. The others are provided in case of desperation.

  1. Use the itk-wasm dicom, package, which is available from here. It has the broadest support and is the easiest to use.

  2. Use the itk::ImageSeriesReader in combination with the DicomSeriesFileNames. For a full example on how to do this, please look at the DicomSeriesReadImageWrite2.cxx code.

  3. Write a MetaImage header. This is a small text file holding information about the image: spacing, dimensions, pixelt type, etc. This header can hold the list of DICOM files you want to read. The only restriction is that the files must be uncompressed. You can take an existing MetaImage header and modify it in order to fit your needs.

  4. Use MRIConvert: Jolinda Smith from the Lewis Center for Neuroimaging at the University of Oregon developed a nice application that allows you to load DICOM series and export them in MetaImage and Analyze format (among others). She graciously has made this application publicly available at: https://darkwing.uoregon.edu/~jolinda/MRIConvert/.

  5. Jeff Brubaker and Stephen R. Aylward, at the UNC CADDLab developed a DICOM query/move application called “MIND” (with the motto: “MIND is not DICOM”). This application loads DICOM files over the network and export them in MetaImage format. This application is open source and it is available at: http://www.jeffro.net/mind/ and http://caddlab.rad.unc.edu/software/MIND/.

How do I write a volume as DICOM series?

Use ITK in combination with GDCM: http://gdcm.sourceforge.net/.

GDCM is an open source package developed by the Creatis team at INSA-Lyon. It is distributed under a license similar to ITK: http://gdcm.sourceforge.net/Copyright.html.

GCDM uses CMake in order to configure its build process, so you will find a familiar setup. Once you download, configure and buid GDCM, you can reconfigure ITK by running CMake in your ITK build, going into the advanced options and enabling USE_GDCM.

For a code example on how to use GDCM for reading and writing DICOM series, please refer to the source code in DicomSeriesReadImageWrite2.cxx.

You can always use the latest and greatest of GDCM, simply use an installed version of GDCM and link ITK to it using ITK_USE_SYSTEM_GDCM.

Which interpreted languages are supported by ITK?

ITK supports a number of language bindings, including Python and JavaScript, among others, through SimpleITK and ITK-Wasm.

How do I cite the use of ITK in a publication?

The Insight software may be cited by referencing the paper, the books, and/or the web site.

  • The papers

McCormick M, Liu X, Jomier J, Marion C, Ibanez L. ITK: enabling
reproducible research and open science. Front Neuroinform. 2014;8:13.
Published 2014 Feb 20. <doi:10.3389/fninf.2014.00013>
T.S. Yoo, M. J. Ackerman, W. E. Lorensen, W. Schroeder, V. Chalana, S.
Aylward, D. Metaxas, R. Whitaker. Engineering and Algorithm Design for
an Image Processing API: A Technical Report on ITK - The Insight
Toolkit. In Proc. of Medicine Meets Virtual Reality, J. Westwood, ed.,
IOS Press Amsterdam pp 586-592 (2002).
  • The books

"The ITK Software Guide: Design and Functionality"
 Fourth Edition
 Johnson, McCormick, Ibanez.
 published by Kitware Inc.
 2015
 ISBN: 9781-930934-28-3
"The ITK Software Guide: Introduction and Development Guidelines"
 Fourth Edition
 Johnson, McCormick, Ibanez.
 published by Kitware Inc.
 2015
 ISBN: 9781-930934-27-6
"Insight into Images"
 edited by Terry Yoo
 published by A.K. Peters
 2004
 ISBN: 1-56881-217-5
"The ITK Software Guide"
 Second Edition
 Ibanez, Schroeder, Ng, Cates.
 published by Kitware Inc.
 2005
 ISBN: 1-930934-15-7
"The ITK Software Guide"
 First Edition
 Ibanez, Schroeder, Ng, Cates.
 published by Kitware Inc.
 2003
 ISBN: 1-930934-10-6
"Mastering CMake, A Cross-Platform Build System"
 K. Martin and B. Hoffman
 published by Kitware Inc.
 ISBN:  1-930934-09-2

BibTeX citations

@manual{johnson2015itk1,
author = {Hans J. Johnson and Matthew M .McCormick and Luis Ib\'{a}{\~n}ez},
title = {The ITK Software Guide Book 1: Introduction and Development Guidelines - Volume 1},
publisher = {Kitware, Inc.},
year = {2015}
}
@manual{johnson2015itk2,
author = {Hans J. Johnson and Matthew M .McCormick and Luis Ib\'{a}{\~n}ez},
title = {The ITK Software Guide Book 2: Design and Functionality - Volume 2},
publisher = {Kitware, Inc.},
year = {2015}
}
@manual{ITKSoftwareGuide,
author = {Luis Ib\'{a}{\~n}ez and William Schroeder and Lydia Ng and Joshua Cates},
title = {The {ITK} {S}oftware {G}uide},
organization = {Kitware, Inc.},
edition = {First},
year = {2003},
note = {ISBN 1-930934-10-6},
url = {}
}
@manual{ITKSoftwareGuideSecondEdition,
author = {Luis Ib\'{a}{\~n}ez and William Schroeder and Lydia Ng and Joshua Cates},
title = {The {ITK} {S}oftware {G}uide},
organization = {Kitware, Inc.},
edition = {Second},
year = {2005},
note = {ISBN 1-930934-15-7},
url = {}
}
@manual{ITKSoftwareGuideThirdEdition,
author = {Hans J. Johnson and Matthew M .McCormick and Luis Ib\'{a}{\~n}ez and The Insight Software Consortium},
title = {The {ITK} {S}oftware {G}uide},
organization = {Kitware, Inc.},
edition = {Third},
year = {2013},
note = {},
url = {}
}
  • A specific software version

See https://zenodo.org/record/3592082.

If you have a publication that used ITK, please create a pull request to add it to ITKBibliography.bib.

If you want to include ITK in an acknowledgment section, a phrase similar to the following may be used:

"This work benefited from the use of the Insight Segmentation
 and Registration Toolkit (ITK), an open source software
 developed as an initiative of the U.S. National Library
 of Medicine and available at www.itk.org."

and

"The multi-platform configuration tool CMake was used for
 configuring ITK and facilitating its use from our project.
 CMake was partially funded by the U.S. National Library of
 Medicine as part of the Insight Toolkit project. CMake is an
 open source system and it is freely available at www.cmake.org."

Is ITK FDA-Approved?

Given the fact that ITK is a software toolkit, it cannot be the subject of FDA approval as a medical device. We have discussed this topic in several occasions and received advice from FDA representatives, that can be summarized as follows:

"ITK is to be considered as an off-the-shelf (OTS) product that
 is used for supporting a higher level medical application/product.
 The developer of such application/product will be responsible for
 performing the validation processes described in FDA published
 guidelines for the development of software-related medical devices."

For mode details see the page FDA guidelines for software development.

Using ITK

How do I configure Vim for ITK coding style?

ITK provides a clang-format configuration file that ensures that the code being typed follows the ITK coding style.

It suffices to tell Vim to use such configuration file.

Contribute

How do I make code contributions to ITK?

Please see the Contribute to ITK page.

Testing

Is ITK tested?

Please see this page: ITK/Testing.

Working with image data

How do I iterate through all the pixels in an image?

Please see the ImageRegionIterator.cxx example.

What are Hounsfield Units?

CT imaging data are measured in Hounsfield Units (HU), which is a quantitative scale for describing radiodensity.

Hounsfield units were defined by Sir Godfrey N. Hounsfield, one of the pioneers of Computer Tomography for clinical applications.

The units represent the linear attenuation of X-Rays for a particular material.

The units scale is defined in such a way that level for Water is 0 and the level for Air is -1000.

The attenuation of any other material is mapped linearly into this range.

The table below shows the HU of various human tissues.

Tissue

HU

Bone

1000

Liver

[40, 60]

White matter

46

Grey matter

43

Blood

40

Muscle

[10, 40]

Kidney

30

Cerebrospinal fluid

15

Water

0

Fat

[-100, -50]

Air

-1000

More details can be found in the Wikipedia article Houndfield Units.

What are MRI T1 / T2 units?

MRI images measure two types of relaxation times, T1 and T2.

The units are millisecons and the two relaxation concepts relate to how long it takes for the molecules in the tissue to realign themselves with a magentic field after being perturbed.

Details on the MRI units are available on the Relaxation (NMR) Wikipedia article.

DICOM: Bits Allocated, Bits Stored and High Bit

A question that is often asked on insight-users is: I am trying to read/write a DICOM image, but some of the DICOM fields have changed after I write it back on disk.

Here is what you have on disk:

Bits Allocated = 16
Bits Stored    = 12
High Bit       = 11

                  |<------------------ pixel ----------------->|
    ______________ ______________ ______________ ______________
   |XXXXXXXXXXXXXX|              |              |              |
   |______________|______________|______________|______________|
    15          12 11           8 7            4 3            0

Here is what you generally get after writing by GDCM:

Bits Allocated = 16
Bits Stored    = 16
High Bit       = 15

   |<------------------------- pixel ------------------------->|
    ______________ ______________ ______________ ______________
   |              |              |              |              |
   |______________|______________|______________|______________|
    15          12 11           8 7            4 3            0

Since DICOM V3, you cannot store any overlay in the unused bits of the Pixel Data (0x7fe0,0x0010), thus it imply that the unused bits are zeros. In ITK, there is no such thing as 12-bit pixel type, thus when instanciating your reader you declare a 16-bit pixel type. This buffer is then passed back to GDCM which sees an array of 16bits values and then simply used the simpliest way to store this buffer back on disk: Bits Stored = 16 and High Bit = 15. In case you have not made any change to your pixel buffer, implicitly the first 4 bits will be zero again. The only difference being that they are not as clearly declared as 12 bits.

I’m working with DICOM images, where can I find more information?

See the DICOM page.

Data display

What imaging software is available for displaying medical image files?

Please see the third party applications page for visualization applications that used ITK to perform image reading/writing.

When I view the output of my program, all I see is a black image. What went wrong?

ITK is capable of reading and writing data in a variety of formats and bit depths. However many “consumer” image programs only work with 8-bit data, so when you try to display a 16-bit data file the image will appear black. The solution is to use an image display program that correctly handles these types of files. A small selection is listed above.

ImageMagick is a particularly useful set of tools for image display, manipulation and conversion.

itkwidgets is capable of correctly loading and displaying all file types supported by ITK (including anisotropic images).

Registration

Can ITK be used for multi-modal registration? For example, CT-MRI, or MRI-fMRI?

ITK provides several method for performing multi-modality registration. These methods are described in detail in the “Registration” chapter of The ITK Software Guide. In particular, you will find interesting the section where the mutual information metric is described.

Examples on multimodality registration are available in Examples/RegistrationITKv4.

It is strongly recommended that you read the chapter on “Geometric Transformations” section, where resampling is explained, before you get involved with image registration methods. That will save you a lot of misunderstandings that are common in new users.

You will also find useful the tutorial sessions, in particular the overview.

Common problems

Why can’t I read any image files? or Why am I getting the exception message: There are no registered IO factories.?

When trying to read or write a file using the itk::ImageFileReader or the itk::ImageFileWriter I got an exception with the following message:

This exception is thrown when ITK does not recognize the fileformat that you are trying to read or write. When reading, ITK recognize file formats by asking a set of internal readers to attempt to read your file. If none of the registered readers accept your file, then the exception above is thrown. When writing, ITK uses the extension of the filename in order to chose the fileformat to be used. Make sure that you use one of the filename extensions recognized by ITK, and that the ImageIO class for that format has been registered.

Enable all the IO modules that you wish to support in your ITK CMake configuration.

To register the IO factory:

  • If you are using CMake: call include(${ITK_USE_FILE}) before add_executable or add_library.

  • If you are not using CMake: the factories should be registered with itk::XXXImageIOFactory::RegisterOneFactory() where XXX is the format.

  • You may need to add something to the REQUIRED list of the find_package call that includes ITK from your project. For example, if you were trying to open a PNG file when you got this error, ensure your CMakeLists.txt contains find_package(ITK REQUIRED ITKIOPNG).