Difference between revisions of "CALAMITI"
Jump to navigation
Jump to search
| (5 intermediate revisions by 2 users not shown) | |||
| Line 1: | Line 1: | ||
| − | <meta name="title" content="CALAMITI"/> | + | <!-- <meta name="title" content="CALAMITI"/> --> |
{{h2|Contrast Anatomy Learning and Analysis for MR Intensity Translation and Integration (CALAMITI)}} | {{h2|Contrast Anatomy Learning and Analysis for MR Intensity Translation and Integration (CALAMITI)}} | ||
{{TOCright}} | {{TOCright}} | ||
| − | Contrast Anatomy Learning and Analysis for MR Intensity Translation and Integration (CALAMITI) is our current MR harmonization method. It was designed to achieve unsupervised multi-site MR harmonization. The associated | + | Contrast Anatomy Learning and Analysis for MR Intensity Translation and Integration (CALAMITI) is our current MR harmonization method. It was designed to achieve unsupervised multi-site MR harmonization. The associated publications are: |
*{{pub| author = L. Zuo, B.E. Dewey, A. Carass, Y. Liu, Y. He, P.A. Calabresi, and J.L. Prince| title = [https://arxiv.org/abs/2103.13283 Information-based Disentangled Representation Learning for Unsupervised MR Harmonization]| conf = ipmi2021}} | *{{pub| author = L. Zuo, B.E. Dewey, A. Carass, Y. Liu, Y. He, P.A. Calabresi, and J.L. Prince| title = [https://arxiv.org/abs/2103.13283 Information-based Disentangled Representation Learning for Unsupervised MR Harmonization]| conf = ipmi2021}} | ||
| − | *{{pub| author= L. Zuo, B. E. Dewey, Y. Liu, Y. He, S. D. Newsome, E. M. Mowry, S. M. Resnick, J. L. Prince, and A. Carass| title = | + | *{{pub| author= L. Zuo, B. E. Dewey, Y. Liu, Y. He, S. D. Newsome, E. M. Mowry, S. M. Resnick, J. L. Prince, and A. Carass| title = Unsupervised MR harmonization by learning disentangled representations using information bottleneck theory| jrnl=ni | number= 118569 | when=2021| doi=10.1016/j.neuroimage.2021.118569}} |
{{h3|Software}} | {{h3|Software}} | ||
| Line 14: | Line 14: | ||
| 220k | | 220k | ||
|- | |- | ||
| − | | CALAMITI (3D) | + | | {{iacl|~lianrui/calamiti_3d/neuroimage_2021_calamiti.zip|CALAMITI (3D)}} |
| − | | | + | | 330Mb (including sample data) |
| | | | ||
|- | |- | ||
| Line 47: | Line 47: | ||
{{h4|Sample command and dependencies}} | {{h4|Sample command and dependencies}} | ||
| − | * After downloading the code, sample command for CALAMITI training | + | * After downloading the code, sample command for CALAMITI training can be found in <code>"script"</code> folder. Testing (encoding and decoding) code can be found in <code>"encode"</code> and <code>"decode"</code>. |
* Conda environment can be downloaded {{iacl|~lianrui/calamiti/calamiti.yml|here}}. | * Conda environment can be downloaded {{iacl|~lianrui/calamiti/calamiti.yml|here}}. | ||
If you have other questions regarding the method or software, please email Lianrui Zuo at <code>lr_zuo@jhu.edu | If you have other questions regarding the method or software, please email Lianrui Zuo at <code>lr_zuo@jhu.edu | ||
Latest revision as of 19:02, 1 August 2022
Contrast Anatomy Learning and Analysis for MR Intensity Translation and Integration (CALAMITI)
Contrast Anatomy Learning and Analysis for MR Intensity Translation and Integration (CALAMITI) is our current MR harmonization method. It was designed to achieve unsupervised multi-site MR harmonization. The associated publications are:
- L. Zuo, B.E. Dewey, A. Carass, Y. Liu, Y. He, P.A. Calabresi, and J.L. Prince, "Information-based Disentangled Representation Learning for Unsupervised MR Harmonization", 27th Conference on Information Processing in Medical Imaging (IPMI 2021), Virtually in Bornholm, Denmark, June 28 - July 2, 2021.
- L. Zuo, B. E. Dewey, Y. Liu, Y. He, S. D. Newsome, E. M. Mowry, S. M. Resnick, J. L. Prince, and A. Carass, "Unsupervised MR harmonization by learning disentangled representations using information bottleneck theory", NeuroImage, 118569, 2021. (doi)
Software
| CALAMITI (2D) | 220k | |
| CALAMITI (3D) | 330Mb (including sample data) |
Instructions
Preprocessing
CALAMITI requires the following preprocessing steps:
- N4 inhomogeneity correction.
- Registration to MNI space (0.8mm isotropic resolution is ideal. Image dimension after MNI registration is 241*288*241).
- White matter peak normalization (see https://github.com/jcreinhold/intensity-normalization).
Prepare training
CALAMITI requires paired multi-contrast MR images (e.g., T1-w and T2-w) during training. The ideal structure of the data directory and naming convention are as follows:
├──absolute_path_to_data
├──SiteA
| ├──train
| | ├──SiteA_T1w_SUB*_ORIENTATION_SLICE*.nii.gz ("ORIENTATION" should be "AXIAL", "CORONAL", or "SAGITTAL")
| | └──SiteA_T2w_SUB*_ORIENTATION_SLICE*.nii.gz
| └──valid
| ├──SiteA_T1w_SUB*_ORIENTATION_SLICE*.nii.gz
| └──SiteA_T2w_SUB*_ORIENTATION_SLICE*.nii.gz
└──SiteB
├──train
| ├──SiteB_T1w_SUB*_ORIENTATION_SLICE*.nii.gz
| └──SiteB_T2w_SUB*_ORIENTATION_SLICE*.nii.gz
└──valid
├──SiteB_T1w_SUB*_ORIENTATION_SLICE*.nii.gz
└──SiteB_T2w_SUB*_ORIENTATION_SLICE*.nii.gz
Sample command and dependencies
- After downloading the code, sample command for CALAMITI training can be found in
"script"folder. Testing (encoding and decoding) code can be found in"encode"and"decode". - Conda environment can be downloaded here.
If you have other questions regarding the method or software, please email Lianrui Zuo at lr_zuo@jhu.edu
