Model-based image synthesize of CT and MR brain images.
This is code for model-based image synthesize (translation) of brain CTs and MRIs. Given any combination of valid input modalities - MR-T1w, MR-T2w, MR-PDw, CT - the missing modalities will be synthesized. For example, if a subject has only a T1w scan, the CT, PDw and T2w scans will be synthesized:
The implementation is done in MATLAB and depends on the SPM12 package (and its MB toolbox). The code should work on raw images; that is, with minimal preprocessing applied beforehand (the input scans just need to be aligned/registered). It furthermore requires no training, as it is fully unsupervised. If you find the code useful, please consider citing the publication in the Reference section.
The algorithm is developed using MATLAB and relies on external functionality from the SPM12 software. The following are therefore required downloads and need to be placed on the MATLAB search path (using addpath):
This example code synthesize CT, PDw and T2w scans from an input T1w MRI. The ouputs are written in the odir folder, prefixed mi*. Note that only the missing modalities are synthesized, others remain the same as their input (but intensity non-uniformity corrected).
files = 'T1w.nii'; % path to T1w MR imagemodalities = 't1'; % inform algorithm that the image is a T1wodir = 'synthesized'; % folder where to write outputspm_synthesize_mri_ct(files, modalities, odir);
This example code synthesize PDw and CT scans from input T1w and T2w MRIs. The ouputs are written in the same folder as the input images, prefixed mi*.
files = {'T1w.nii', 'T2w.nii'}; % path to PDw MR imagemodalities = {'t1', 't2'}; % inform algorithm that the image is a PDwspm_synthesize_mri_ct(files, modalities);
@inproceedings{brudfors2019empirical, title={Empirical bayesian mixture models for medical image translation}, author={Brudfors, Mikael and Ashburner, John and Nachev, Parashkev and Balbastre, Ya{\"e}l}, booktitle={International Workshop on Simulation and Synthesis in Medical Imaging}, pages={1--12}, year={2019}, organization={Springer} }
This software is free but copyright software, distributed under the terms of the GNU General Public Licence as published by the Free Software Foundation (either version 2, or at your option, any later version).