Repository for project work of Group 4 on 8DM20, Capita Selecta in Medical Image Analysis
#1. Introduction #2. Contents of this package #3. Installation guide #4. Getting started #5. Changing Parameters and Functions
This script package was written for Capita Selecta in Medical Image Analysis (8QC00), a project running in Q3 of academic year 2018-2019 for (B)ME. The scripts were written by group 4 for the case 'Evaluating the Effects of Atlas Filtering and Different Methods of Decision Fusing on Atlas Based Image Segmentation of Prostate Images'.
Members of group 7:
- Leander van Eekelen (0966869, l.v.eekelen@student.tue.nl)
- Kees van Dorp (0653884, a.c.v.dorp@student.tue.nl)
- Merel Eussen (0964211, m.j.a.eussen@student.tue.nl)
- Milan Gillissen (0861128, m.gillissen@student.tue.nl)
- Teun van de Meerakker (0865482 t.v.d.meerakkker@student.tue.nl)
This script package offers the Python scripts necessary to perform Image Registration, Mask Deformation and different descision fusing techniques.
The package the user has received from group 4 should contain the following scripts and parameterfiles.
Scripts in main folder:
- pipeline.py
- Import_Files.py
- Registration.py
- Transform_GT.py
The package 'newmethods', bundling the following three scripts for convenience:
- decisionfusing.py
- Mutualinformation.py
- dice.py
Moreover, it should contain the following parameter files in the folder 'parameterfiles':
- Parameters_BSpline.txt
- Parameters_Affine.txt
If one or more items on this list are missing, please contact one of the group members, mentioned in #1 - Introduction.
To function, the user can clone this github repo to his local machine. The scripts require a Python environment. Assume the most up-to-date Python release is needed. Additionally, the SimpleITK python wrapper of the ITK library is needed.
Furthermore the script uses the elasix and transformix software: S. Klein, M. Staring, K. Murphy, M. A. Viergever and J. P. W. Pluim, "elastix: A Toolbox for Intensity-Based Medical Image Registration," in IEEE Transactions on Medical Imaging, vol. 29, no. 1, pp. 196-205, Jan. 2010. doi: 10.1109/TMI.2009.2035616
To start the script the user should run the 'pipeline.py' Python file. This combines all methods and initiates leave 1-out registration of the entire dataset, given the RUNREGISTRATION parameter is set to True.
If the user wants to purely visualize the data, the RUNREGISTRATION parameter should be set to False. No registration will take place: instead, data will be taken from disk, which is assumed to have already been generated by that point.
Furthermore:
- The Threshold value belonging to the atlas filtering process of Klein et al. has a default value of 0.7
- The following paths should be set to the users own paths
- DATA_PATH: path to the data provided (described in #2). Path should contain the p*** folders.
- ELASTIX_PATH : path to the elasix execute file.
- RESULTS_PATH : path were the resulting data is saved.
- TRANSFORMIX_PATH : path to the transformix execute file.
The process of registering can take a while: expect over an hour of running the script. In the results_path folders structures of fixed_image/moving_images will occur. With:
- IterationInfo.0.R#: The iteration information for all 4 resolutions.
- results.0.mhd: The deformed moving MRI image
- results: The deformed ground truth of the moving MRI image
- TransformParameters.0.R# : The transform parameters for all 4 resolutions.
For each of the voting methods a binary masks will be returned as an mhd file, in the results subfolder: Results_method_# Each folder contains a binary mhd file, consisting of the segmentation of the prostate.
Changes in Registration can be made in de following line, by adding registration = 'Affine', 'BSpline'(default) or 'Both'). MI = register(DATA_PATH, ELASTIX_PATH, RESULT_PATH) To alter the registration parameters, changes can be made in the corresponding parameterfiles: Parameters_BSpline.txt or Parameters_Affine.txt located in the parameterfiles folder.