Modules:MRIBiasFieldCorrection-Documentation-3.6

Return to Slicer 3.6 Documentation

Gallery of New Features

Module Name

MRIBiasFieldCorrection

MRIBiasFieldCorrection corrects for the intensity inhomogeneity in 3D MRI images.

General Information

Module Type & Category

Type: CLI

Category: Filtering

Authors, Collaborators & Contact

Sylvain Jaume: MIT CSAIL Nicolas Rannou: ISEN IRISA Ron Kikinis: Harvard Medical School SPL Contact: sylvain at csail.mit.edu

This work was presented at the MIT NAMIC Week in 2009.

Module Description

MRIBiasFieldCorrection corrects for the intensity inhomogeneity due to MRI gain field distortion. It first estimates the bias field and then 'subtract' it from the image. This module implements the so called methods 'N3' and 'N4' (see references at the bottom of this page).

Usage

Step 1/24. To open the input image, click on 'File' and then 'Add Volume': Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 1/24. Add the input volumes

Step 2/24. Select the input image and click Apply: Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 2/24. Select the input image

Step 3/24. The input image appears in the axial-sagittal-coronal views: Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 3/24. Image displayed in the orthogonal views

Step 4/24. Select the mask image. Check the box 'Label Map' and click Apply: Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 4/24. Select the mask image as a label map

Step 5/24. The label map is overlaid on the axial-sagittal-coronal views: Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 5/24. Label map overlaid on the input image

Step 6/24. Click on Modules and scroll-down the menu: Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 6/24. List of Slicer 3.6 modules

Step 7/24. Select 'Filtering' and then 'MRIBiasFieldCorrection': Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 7/24. Select MRIBiasFieldCorrection in Filtering

Step 8/24. The User Interface of MRIBiasFieldCorrection appears: Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 8/24. User interface of MRIBiasFieldCorrection

Step 9/24. Select the input image in the list of loaded volumes: Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 9/24. Select input among loaded volumes

Step 10/24. Select the mask image in the list of loaded volumes: Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 10/24. Select mask among the loaded label maps

Step 11/24. Create a new node for the output in Slicer MRML tree: Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 11/24. Create node in MRML tree for output

Step 12/24. Click 'Apply' to run the N3 algorithm (default): Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 12/24. Run the N3 algorithm

Step 13/24. 'Status Running' means the algorithm is running. Note 'Running...' in lower bar: Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 13/24. Status info when algorithm is running

Step 14/24. 'Status Completed' means the algorithm is done. Note 'Completed' in lower bar: Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 14/24. Status info when algorithm is done

Step 15/24. : Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 15

Step 16/24. : Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 16

Step 17/24. : Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 17

Step 18/24. : Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 18

Step 19/24. : Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 19

Step 20/24. : Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 20

Step 21/24. : Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 21

Step 22/24. : Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 22

Step 23/24. : Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 23

Step 24/24. : Input Image: input image Mask Image: label map providing an approximative segmentation of the ROI Output Volume: corrected image

Step 24

Screenshot of the MRIBiasFieldCorrection module in Slicer3 during development. The screenshot illustrates the application of the MRI Bias Field Correction on the SPL Atlas. Note that the interface was simplified for the production version. The top left window shows the mask used to defined the ROI where the correction will be applied. The top right window shows the rendering of the 3D model created using the mask. The bottom left window shows one slice in the input image. Note the intensity inhomogeneity from the bottom left corner (dark) to the top right corner (bright). The bottom right window shows the output image after the application of the MRI Bias Field Correction module. The parameters used for the correction are shown on the left panel of the Slicer3 interface.

Cheat Sheet

• Load the input dataset (Main menu: Add Volume)
• Create a Mask Volume using the Editor (Modules > Editor) (the Threshold tool should give a good result)
• Select the MRIBiasFieldCorrection module (Modules > Filtering > MRIBiasFieldCorrection)
• In the left panel, select the Input Volume
• Select the Mask Volume
• In the Preview Volume menu , select 'Create New Volume'
• Do the same for the Output Volume menu
• Modify the parameter values if desired (default values gave good results during our experiments)
• Click on Apply.

Close-up of the above screenshot showing the image before and after correction. Note that the intensity inhomogeneity visible in the input image (left) has been corrected in the output image (right).

It took 32 min to process a 512x512x30 MRI volume on a MacBook laptop. To visualize the result:

• Deselect the Overlay volume,
• Select the input as Background volume and the output as Foreground volume.
• Go to Modules > Volumes
• Select the input volume, and write down the values for Window and Level
• Select the output volume, and apply the same values
• Move the lower left cursor in 'Manipulate Slice Views' between B (background) and F (foreground)

Tests

On the Dashboard, these tests verify that the module is working on various platforms:

• MyModuleTest1 MyModuleTest1.cxx
• MyModuleTest2 MyModuleTest2.cxx

Known bugs

Links to known bugs in the Slicer3 bug tracker

• Bug 000: description

Usability issues

Follow this link to the Slicer3 bug tracker. Please select the usability issue category when browsing or contributing.

Source code & documentation

Links to the module's source code:

Source code:

• Source code

Doxygen documentation:

• class1

More Information

Acknowledgment

This work is part of the National Alliance for Medical Image Computing (NAMIC), funded by the National Institutes of Health through the NIH Roadmap for Medical Research, Grant U54 EB005149 (PI: Ron Kikinis).

References for algorithms implemented in this module

• A Nonparametric Method for Automatic Correction of Intensity Nonuniformity in MRI Data, J.G. Sled, A.P. Zijdenbos, and A.C. Evans, IEEE Transactions on Medical Imaging, 17(1):87–97, Feb 1998.
• N4ITK: Improved N3 Bias Correction, N.J. Tustison, B.B. Avants, P.A. Cook, Y. Zheng, A. Egan, P.A. Yushkevich, J.C. Gee, IEEE Transactions on Medical Imaging, Vol 99, April 2010.
• N4ITK: Nick's N3 ITK Implementation for MRI Bias Field Correction, N. Tustison, J. Gee, Insight Journal, 2009.

References for related algorithms

• Parametric Estimate of Intensity Inhomogeneities Applied to MRI, M. Styner, C. Brechbhuler, G. Szekely, and G. Gerig, IEEE Transactions on Medical Imaging, 19(3):153–165, Mar 2000.