Skip to content

Latest commit

 

History

History
472 lines (409 loc) · 32.3 KB

README.md

File metadata and controls

472 lines (409 loc) · 32.3 KB
Raw

An active texture-based digital atlas enables automated mapping of structures and markers across brains. Yuncong Chen, Lauren E. McElvain, Alexander S. Tolpygo, Daniel Ferrante, Beth Friedman, Partha P. Mitra, Harvey J. Karten, Yoav Freund & David Kleinfeld. Nature Methods (2019/3/11)

This toolkit is written in Python 2.7.2 and has been tested on a machine with Intel Xeon W5580 3.20GHz 16-core CPU, 128GB RAM and a Nvidia Titan X GPU, running Linux Ubuntu 16.04.

A complete run-through of the following demo takes roughly 2 hours. In order to allow an user to walk through the pipeline in as little time as possible, this demo only used 3 sections. Such small number of sections is not adequate to obtain an accurate registration. To get reasonable results, at least 10 sections are recommended for each landmark.

Associated Neuroglancer viewer of the three foundational thionin brains available here.

Installation

Install CUDA (refer to this page)

wget https://developer.nvidia.com/compute/cuda/9.0/Prod/local_installers/cuda_9.0.176_384.81_linux-run`
sudo chmod +x cuda_9.0.176_384.81_linux-run
sudo ./cuda_9.0.176_384.81_linux-run
  • Select "no" to “Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 384.81?”.
  • Then download cuDNN (latest version for CUDA 9.0)
tar xvzf cudnn-9.0-linux-x64-v7.4.2.24.tgz
sudo cp -P cuda/include/cudnn.h /usr/local/cuda/include
sudo cp -P cuda/lib64/libcudnn* /usr/local/cuda/lib64
sudo chmod a+r /usr/local/cuda/include/cudnn.h /usr/local/cuda/lib64/libcudnn*
sudo ldconfig

Install other non-python packages

  • Install ImageMagick 6.8.9. sudo apt-get install imagemagick
  • To use GUIs, install PyQt4 into the virtualenv according to this answer.
    • Install python-qt4 globaly: sudo apt-get install python-qt4
    • Create symbolic link of PyQt4 to your virtual env: ln -s /usr/lib/python2.7/dist-packages/PyQt4/ mousebrainatlas_virtualenv/lib/python2.7/site-packages/
    • Create symbolic link of sip.so to your virtual env: ln -s /usr/lib/python2.7/dist-packages/sip.x86_64-linux-gnu.so mousebrainatlas_virtualenv/lib/python2.7/site-packages/

Install python packages

A configuration script is provided to create a virtualenv called mousebrainatlas-virtualenv and install necessary packages.

  • Change REPO_DIR, ROOT_DIR, DATA_ROOTDIR, THUMBNAIL_DATA_ROOTDIR in setup/config.sh
  • The default requirements.txt assumes CUDA version of 9.0. If your CUDA version (check using nvcc -V or cat /usr/local/cuda/version.txt) is 9.1, replace mxnet-cu90 with mxnet-cu91 in requirements.txt. If your machine does not have a GPU, replace mxnet-cu90 with mxnet. Refer to official mxnet page for available pips.
  • source setup/config.sh. Make sure we are now working under the mousebrainatlas_virtualenv virtual environment.
  • cd demo.

Preprocessing

Note that the DEMO998_input_spec.ini files for most steps are different and must be manually created according to the actual input. In the following instructions, "create DEMO998_input_spec.ini as (prep_id, version, resolution)" means using the same set of image names as image_name_list but set the prep_id, version and resolution accordingly.

  • Download demo data. Run python download_demo_data.py to download necessary data.

Make sure the folder content looks like:

├── brains_info
│   └── DEMO998.ini
├── jp2_files
│   └── DEMO998
│       ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_lossless.jp2
│       ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_lossless.jp2
│       └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_lossless.jp2
├── mxnet_models
│   └── inception-bn-blue
│       ├── inception-bn-blue-0000.params
│       ├── inception-bn-blue-symbol.json
│       └── mean_224.npy
└── CSHL_data_processed
│   └── DEMO998
│       └── DEMO998_sorted_filenames.txt
│       └── DEMO998_prep2_sectionLimits
└── operation_configs
    ├── crop_orig_template.ini
    ├── from_aligned_to_none.ini
    ├── from_aligned_to_padded.ini
    ├── from_none_to_aligned_template.ini
    ├── from_none_to_padded.ini
    ├── from_none_to_wholeslice.ini
    ├── from_padded_to_none.ini
    ├── from_padded_to_brainstem_template.ini
    ├── from_padded_to_wholeslice_template.ini
    └── from_wholeslice_to_brainstem.ini
  • Convert raw images from JPEG2000 to tif. Edit DEMO998_raw_input_spec.json. Set data_dirs, filepath_to_imageName_mapping and imageName_to_filepath_mapping. Run python jp2_to_tiff.py DEMO998 DEMO998_raw_input_spec.json.
├── CSHL_data_processed
│   └── DEMO998
│       ├── DEMO998_raw
│       │   ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_raw.tif
│       │   ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_raw.tif
│       │   └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_raw.tif
  • Extract Neurotrace-blue channel. Modify DEMO998_input_spec.ini as (None,None,raw). python extract_channel.py example_specs/DEMO998_input_spec.ini 2 Ntb
├── CSHL_data_processed
│   └── DEMO998
│       └── DEMO998_raw_Ntb
│           ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_raw_Ntb.tif
│           ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_raw_Ntb.tif
│           └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_raw_Ntb.tif
  • Rescale to thumbnail. Modify DEMO998_input_spec.ini as (None,Ntb,raw). python rescale.py example_specs/DEMO998_input_spec.ini thumbnail -f 0.03125
├── CSHL_data_processed
│   └── DEMO998
│       └── DEMO998_thumbnail_Ntb
│           ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_thumbnail_Ntb.tif
│           ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_thumbnail_Ntb.tif
│           └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_thumbnail_Ntb.tif
  • Global intensity normalization. Modify DEMO998_input_spec.ini as (None,Ntb,thumbnail). python normalize_intensity.py example_specs/DEMO998_input_spec.ini NtbNormalized
├── CSHL_data_processed
│   └── DEMO998
│       └── DEMO998_thumbnail_NtbNormalized
│           ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_thumbnail_NtbNormalized.tif
│           ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_thumbnail_NtbNormalized.tif
│           └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_thumbnail_NtbNormalized.tif
  • Create an ordered list of images. Create $DATA_ROOTDIR/CSHL_data_processed/DEMO998/DEMO998_sorted_filenames.txt. This file should already be included in the initial download. Each row of the file contains an image name and its index. The file should look like:
MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242 225
MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250 230
MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257 235
  • Align images in this stack.
    • Copy operation config template cp operation_configs/from_none_to_aligned_template.ini CSHL_data_processed/DEMO998/DEMO998_operation_configs/from_none_to_aligned.ini. Modify from_none_to_aligned.ini. In particular make sure elastix_parameter_fp is valid.
    • Modify DEMO998_input_spec.ini as (None,NtbNormalized,thumbnail). python align_compose.py example_specs/DEMO998_input_spec.ini --op from_none_to_aligned
├── CSHL_data_processed
│   └── DEMO998
│       └── DEMO998_transformsTo_MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250.csv
│       ├── DEMO998_elastix_output
│       │   ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_to_MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242
│       │   │   ├── elastix.log
│       │   │   ├── IterationInfo.0.R0.txt
│       │   │   ├── IterationInfo.0.R1.txt
│       │   │   ├── IterationInfo.0.R2.txt
│       │   │   ├── IterationInfo.0.R3.txt
│       │   │   ├── IterationInfo.0.R4.txt
│       │   │   ├── IterationInfo.0.R5.txt
│       │   │   ├── result.0.tif
│       │   │   └── TransformParameters.0.txt
│       │   └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_to_MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250
│       │       ├── elastix.log
│       │       ├── IterationInfo.0.R0.txt
│       │       ├── IterationInfo.0.R1.txt
│       │       ├── IterationInfo.0.R2.txt
│       │       ├── IterationInfo.0.R3.txt
│       │       ├── IterationInfo.0.R4.txt
│       │       ├── IterationInfo.0.R5.txt
│       │       ├── result.0.tif
│       │       └── TransformParameters.0.txt
  • Transform images. python warp_crop.py --input_spec example_specs/DEMO998_input_spec.ini --op_id from_none_to_padded --njobs 8
├── CSHL_data_processed
│   └── DEMO998
│       ├── DEMO998_prep1_thumbnail_NtbNormalized
│       │   ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_prep1_thumbnail_NtbNormalized.tif
│       │   ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_prep1_thumbnail_NtbNormalized.tif
│       │   └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_prep1_thumbnail_NtbNormalized.tif

  • Make sure the all_stacks meta-variable in src/utilities/metadata.py includes DEMO998.

  • Generate masks. The masks should be included in the initial download. For how to generate them from scratch. refer to this page.

├── CSHL_data_processed
│   └── DEMO998
│       ├── DEMO998_prep1_thumbnail_mask
│       │   ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_prep1_thumbnail_mask.png
│       │   ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_prep1_thumbnail_mask.png
│       │   └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_prep1_thumbnail_mask.png
│       ├── DEMO998_thumbnail_mask
│       │   ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_thumbnail_mask.png
│       │   ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_thumbnail_mask.png
│       │   └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_thumbnail_mask.png
  • Local adaptive intensity normalization. Modify DEMO998_input_spec.ini as (None,Ntb,raw). python normalize_intensity_adaptive.py input_spec.ini NtbNormalizedAdaptiveInvertedGamma
├── CSHL_data_processed
│   └── DEMO998
│       ├── DEMO998_intensity_normalization_results
│       │   ├── floatHistogram
│       │   │   ├── DEMO998_MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_raw_floatHistogram.png
│       │   │   ├── DEMO998_MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_raw_floatHistogram.png
│       │   │   └── DEMO998_MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_raw_floatHistogram.png
│       │   ├── meanMap
│       │   │   ├── DEMO998_MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_raw_meanMap.bp
│       │   │   ├── DEMO998_MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_raw_meanMap.bp
│       │   │   └── DEMO998_MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_raw_meanMap.bp
│       │   ├── meanStdAllRegions
│       │   │   ├── DEMO998_MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_raw_meanStdAllRegions.bp
│       │   │   ├── DEMO998_MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_raw_meanStdAllRegions.bp
│       │   │   └── DEMO998_MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_raw_meanStdAllRegions.bp
│       │   ├── normalizedFloatMap
│       │   │   ├── DEMO998_MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_raw_normalizedFloatMap.bp
│       │   │   ├── DEMO998_MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_raw_normalizedFloatMap.bp
│       │   │   └── DEMO998_MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_raw_normalizedFloatMap.bp
│       │   ├── regionCenters
│       │   │   ├── DEMO998_MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_raw_regionCenters.bp
│       │   │   ├── DEMO998_MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_raw_regionCenters.bp
│       │   │   └── DEMO998_MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_raw_regionCenters.bp
│       │   └── stdMap
│       │       ├── DEMO998_MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_raw_stdMap.bp
│       │       ├── DEMO998_MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_raw_stdMap.bp
│       │       └── DEMO998_MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_raw_stdMap.bp
  • Whole-slice crop.
    • Copy operation config template cp $DATA_ROOTDIR/operation_configs/from_padded_to_wholeslice_template.ini $DATA_ROOTDIR/CSHL_data_processed/DEMO998/DEMO998_operation_configs/from_padded_to_wholeslice.ini. Modify from_padded_to_wholeslice.ini. In this file specify the cropbox for the domain alignedWithMargin based on alignedPadded images.
    • Modify DEMO998_input_spec.ini as (None,NtbNormalizedAdaptiveInvertedGamma,raw). python warp_crop.py --input_spec example_specs/DEMO998_input_spec.ini --op_id from_none_to_wholeslice
├── CSHL_data_processed
│   └── DEMO998
│       ├── DEMO998_prep5_raw_NtbNormalizedAdaptiveInvertedGamma
│       │   ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_prep5_raw_NtbNormalizedAdaptiveInvertedGamma.tif
│       │   ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_prep5_raw_NtbNormalizedAdaptiveInvertedGamma.tif
│       │   └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_prep5_raw_NtbNormalizedAdaptiveInvertedGamma.tif
  • Modify input_spec.ini as (alignedWithMargin,NtbNormalizedAdaptiveInvertedGamma,raw). python rescale.py example_specs/DEMO998_input_spec.ini thumbnail -f 0.03125
├── CSHL_data_processed
│   └── DEMO998
│       ├── DEMO998_prep5_thumbnail_NtbNormalizedAdaptiveInvertedGamma
│       │   ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_prep5_thumbnail_NtbNormalizedAdaptiveInvertedGamma.tif
│       │   ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_prep5_thumbnail_NtbNormalizedAdaptiveInvertedGamma.tif
│       │   └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_prep5_thumbnail_NtbNormalizedAdaptiveInvertedGamma.tif
  • Brainstem crop.
    • Copy operation config template cp operation_configs/from_padded_to_brainstem_template.ini CSHL_data_processed/DEMO998/DEMO998_operation_configs/from_padded_to_brainstem.ini. Modify from_padded_to_brainstem.ini.
    • Modify DEMO998_input_spec.ini as (alignedWithMargin,NtbNormalizedAdaptiveInvertedGamma,raw). python warp_crop.py --input_spec example_specs/DEMO998_input_spec.ini --op_id from_wholeslice_to_brainstem
├── CSHL_data_processed
│   └── DEMO998
│       ├── DEMO998_prep2_raw_NtbNormalizedAdaptiveInvertedGamma
│       │   ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_prep2_raw_NtbNormalizedAdaptiveInvertedGamma.tif
│       │   ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_prep2_raw_NtbNormalizedAdaptiveInvertedGamma.tif
│       │   └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_prep2_raw_NtbNormalizedAdaptiveInvertedGamma.tif
  • Generate thumbnails. Modify DEMO998_input_spec.ini as (alignedBrainstemCrop, NtbNormalizedAdaptiveInvertedGamma, raw). python rescale.py example_specs/DEMO998_input_spec.ini thumbnail -f 0.03125
├── CSHL_data_processed
│   └── DEMO998
│       ├── DEMO998_prep2_thumbnail_NtbNormalizedAdaptiveInvertedGamma
│       │   ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_prep2_thumbnail_NtbNormalizedAdaptiveInvertedGamma.tif
│       │   ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_prep2_thumbnail_NtbNormalizedAdaptiveInvertedGamma.tif
│       │   └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_prep2_thumbnail_NtbNormalizedAdaptiveInvertedGamma.tif
  • Compress JPEG. Use the same DEMO998_input_spec.ini as previous step. python compress_jpeg.py example_specs/DEMO998_input_spec.ini
├── CSHL_data_processed
│   └── DEMO998
│       ├── DEMO998_prep2_raw_NtbNormalizedAdaptiveInvertedGammaJpeg
│       │   ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_prep2_raw_NtbNormalizedAdaptiveInvertedGammaJpeg.jpg
│       │   ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_prep2_raw_NtbNormalizedAdaptiveInvertedGammaJpeg.jpg
│       │   └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_prep2_raw_NtbNormalizedAdaptiveInvertedGammaJpeg.jpg
  • Identify section limits that only contain the brainstem. Create $DATA_ROOTDIR/CSHL_data_processed/DEMO998/DEMO998_prep2_sectionLimit.ini. This file should already be included in the initial download. The content looks like:
[DEFAULT]
left_section_limit = 225
right_section_limit = 235

Registration

  • Download atlas. Run python download_atlas.py.
├── CSHL_volumes
│   ├── atlasV7
│   │   └── atlasV7_10.0um_scoreVolume
│   │       └── score_volumes
│   │           ├── atlasV7_10.0um_scoreVolume_12N.bp
│   │           ├── atlasV7_10.0um_scoreVolume_12N_origin_wrt_canonicalAtlasSpace.txt
│   │           ├── atlasV7_10.0um_scoreVolume_12N_surround_200um.bp
│   │           ├── atlasV7_10.0um_scoreVolume_12N_surround_200um_origin_wrt_canonicalAtlasSpace.txt
│   │           ├── atlasV7_10.0um_scoreVolume_3N_R.bp
│   │           ├── atlasV7_10.0um_scoreVolume_3N_R_origin_wrt_canonicalAtlasSpace.txt
│   │           ├── atlasV7_10.0um_scoreVolume_3N_R_surround_200um.bp
│   │           ├── atlasV7_10.0um_scoreVolume_3N_R_surround_200um_origin_wrt_canonicalAtlasSpace.txt
│   │           ├── atlasV7_10.0um_scoreVolume_4N_R.bp
│   │           ├── atlasV7_10.0um_scoreVolume_4N_R_origin_wrt_canonicalAtlasSpace.txt
│   │           ├── atlasV7_10.0um_scoreVolume_4N_R_surround_200um.bp
│   │           └── atlasV7_10.0um_scoreVolume_4N_R_surround_200um_origin_wrt_canonicalAtlasSpace.txt
  • Compute rough global transform. The relevant results should already be included in the initial download. For details on how to obtain it from scratch, follow the instructions on this page.
├── CSHL_simple_global_registration
│   ├── DEMO998_registered_atlas_structures_wrt_wholebrainXYcropped_xysecTwoCorners.json
│   └── DEMO998_T_atlas_wrt_canonicalAtlasSpace_subject_wrt_wholebrain_atlasResol.txt
  • Download pre-trained classifiers. Run python download_pretrained_classifiers.py -s "[\"12N\", \"3N\", \"4N\"]".
├── CSHL_classifiers
│   └── setting_899
│       └── classifiers
│           ├── 12N_clf_setting_899.dump
│           ├── 3N_clf_setting_899.dump
│           └── 4N_clf_setting_899.dump
  • Generate 3-D probability maps. Run python generate_prob_volumes.py DEMO998 799 NtbNormalizedAdaptiveInvertedGamma NtbNormalizedAdaptiveInvertedGammaJpeg -s "[\"12N\", \"3N\", \"4N\"]".
├── CSHL_patch_features
│   └── inception-bn-blue
│       └── DEMO998
│           └── DEMO998_prep2_none_win7
│               ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_prep2_none_win7_inception-bn-blue_features.bp
│               ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_prep2_none_win7_inception-bn-blue_locations.txt
│               ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_prep2_none_win7_inception-bn-blue_features.bp
│               ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_prep2_none_win7_inception-bn-blue_locations.txt
│               ├── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_prep2_none_win7_inception-bn-blue_features.bp
│               └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_prep2_none_win7_inception-bn-blue_locations.txt
├── CSHL_scoremaps
│   └── 10.0um
│       └── DEMO998
│           └── DEMO998_prep2_10.0um_detector799
│               ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_prep2_10.0um_detector799
│               │   ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_prep2_10.0um_detector799_12N_scoremap.bp
│               │   ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_prep2_10.0um_detector799_3N_scoremap.bp
│               │   └── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_prep2_10.0um_detector799_4N_scoremap.bp
│               ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_prep2_10.0um_detector799
│               │   ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_prep2_10.0um_detector799_12N_scoremap.bp
│               │   ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_prep2_10.0um_detector799_3N_scoremap.bp
│               │   └── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_prep2_10.0um_detector799_4N_scoremap.bp
│               └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_prep2_10.0um_detector799
│                   ├── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_prep2_10.0um_detector799_12N_scoremap.bp
│                   ├── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_prep2_10.0um_detector799_3N_scoremap.bp
│                   └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_prep2_10.0um_detector799_4N_scoremap.bp
├── CSHL_scoremap_viz
│   └── 10.0um
│       ├── 12N
│       │   └── DEMO998
│       │       └── detector799
│       │           └── prep2
│       │               ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_prep2_10.0um_12N_detector799_scoremapViz.jpg
│       │               ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_prep2_10.0um_12N_detector799_scoremapViz.jpg
│       │               └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_prep2_10.0um_12N_detector799_scoremapViz.jpg
│       ├── 3N
│       │   └── DEMO998
│       │       └── detector799
│       │           └── prep2
│       │               ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_prep2_10.0um_3N_detector799_scoremapViz.jpg
│       │               ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_prep2_10.0um_3N_detector799_scoremapViz.jpg
│       │               └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_prep2_10.0um_3N_detector799_scoremapViz.jpg
│       └── 4N
│           └── DEMO998
│               └── detector799
│                   └── prep2
│                       ├── MD662&661-F81-2017.06.06-12.44.40_MD661_2_0242_prep2_10.0um_4N_detector799_scoremapViz.jpg
│                       ├── MD662&661-F84-2017.06.06-14.03.51_MD661_1_0250_prep2_10.0um_4N_detector799_scoremapViz.jpg
│                       └── MD662&661-F86-2017.06.06-14.56.48_MD661_2_0257_prep2_10.0um_4N_detector799_scoremapViz.jpg
├── CSHL_volumes
│   └── DEMO998
│       ├── DEMO998_detector799_10.0um_scoreVolume
│       │   ├── score_volume_gradients
│       │   │   ├── DEMO998_detector799_10.0um_scoreVolume_12N_gradients.bp
│       │   │   ├── DEMO998_detector799_10.0um_scoreVolume_12N_origin_wrt_wholebrain.txt
│       │   │   ├── DEMO998_detector799_10.0um_scoreVolume_3N_R_gradients.bp
│       │   │   ├── DEMO998_detector799_10.0um_scoreVolume_3N_R_origin_wrt_wholebrain.txt
│       │   │   ├── DEMO998_detector799_10.0um_scoreVolume_4N_R_gradients.bp
│       │   │   └── DEMO998_detector799_10.0um_scoreVolume_4N_R_origin_wrt_wholebrain.txt
│       │   └── score_volumes
│       │       ├── DEMO998_detector799_10.0um_scoreVolume_12N.bp
│       │       ├── DEMO998_detector799_10.0um_scoreVolume_12N_origin_wrt_wholebrain.txt
│       │       ├── DEMO998_detector799_10.0um_scoreVolume_3N_R.bp
│       │       ├── DEMO998_detector799_10.0um_scoreVolume_3N_R_origin_wrt_wholebrain.txt
│       │       ├── DEMO998_detector799_10.0um_scoreVolume_4N_R.bp
│       │       └── DEMO998_detector799_10.0um_scoreVolume_4N_R_origin_wrt_wholebrain.txt
  • Register 12N (hypoglossal nucleus). Run python register_brains.py example_specs/demo_fixed_brain_spec_12N.json example_specs/demo_moving_brain_spec_12N.json -g.
  • Register 3N(oculomotor nucleus)/4N(trochlear nucleus) complex. Run python register_brains.py example_specs/demo_fixed_brain_spec_3N_R_4N_R.json example_specs/demo_moving_brain_spec_3N_R_4N_R.json -g
├── CSHL_registration_parameters
│   └── atlasV7
│       ├── atlasV7_10.0um_scoreVolume_12N_warp7_DEMO998_detector799_10.0um_scoreVolume_12N
│       │   ├── atlasV7_10.0um_scoreVolume_12N_warp7_DEMO998_detector799_10.0um_scoreVolume_12N_parameters.json
│       │   ├── atlasV7_10.0um_scoreVolume_12N_warp7_DEMO998_detector799_10.0um_scoreVolume_12N_scoreEvolution.png
│       │   ├── atlasV7_10.0um_scoreVolume_12N_warp7_DEMO998_detector799_10.0um_scoreVolume_12N_scoreHistory.bp
│       │   └── atlasV7_10.0um_scoreVolume_12N_warp7_DEMO998_detector799_10.0um_scoreVolume_12N_trajectory.bp
│       └── atlasV7_10.0um_scoreVolume_3N_R_4N_R_warp7_DEMO998_detector799_10.0um_scoreVolume_3N_R_4N_R
│           ├── atlasV7_10.0um_scoreVolume_3N_R_4N_R_warp7_DEMO998_detector799_10.0um_scoreVolume_3N_R_4N_R_parameters.json
│           ├── atlasV7_10.0um_scoreVolume_3N_R_4N_R_warp7_DEMO998_detector799_10.0um_scoreVolume_3N_R_4N_R_scoreEvolution.png
│           ├── atlasV7_10.0um_scoreVolume_3N_R_4N_R_warp7_DEMO998_detector799_10.0um_scoreVolume_3N_R_4N_R_scoreHistory.bp
│           └── atlasV7_10.0um_scoreVolume_3N_R_4N_R_warp7_DEMO998_detector799_10.0um_scoreVolume_3N_R_4N_R_trajectory.bp
├── CSHL_volumes
│   ├── atlasV7
│   │   ├── atlasV7_10.0um_scoreVolume_12N_warp7_DEMO998_detector799_10.0um_scoreVolume_12N_10.0um
│   │   │   └── score_volumes
│   │   │       ├── atlasV7_10.0um_scoreVolume_12N_warp7_DEMO998_detector799_10.0um_scoreVolume_12N_10.0um_12N.bp
│   │   │       ├── atlasV7_10.0um_scoreVolume_12N_warp7_DEMO998_detector799_10.0um_scoreVolume_12N_10.0um_12N_origin_wrt_fixedWholebrain.txt
│   │   │       ├── atlasV7_10.0um_scoreVolume_12N_warp7_DEMO998_detector799_10.0um_scoreVolume_12N_10.0um_12N_surround_200um.bp
│   │   │       └── atlasV7_10.0um_scoreVolume_12N_warp7_DEMO998_detector799_10.0um_scoreVolume_12N_10.0um_12N_surround_200um_origin_wrt_fixedWholebrain.txt
│   │   ├── atlasV7_10.0um_scoreVolume_3N_R_4N_R_warp7_DEMO998_detector799_10.0um_scoreVolume_3N_R_4N_R_10.0um
│   │   │   └── score_volumes
│   │   │       ├── atlasV7_10.0um_scoreVolume_3N_R_4N_R_warp7_DEMO998_detector799_10.0um_scoreVolume_3N_R_4N_R_10.0um_3N_R.bp
│   │   │       ├── atlasV7_10.0um_scoreVolume_3N_R_4N_R_warp7_DEMO998_detector799_10.0um_scoreVolume_3N_R_4N_R_10.0um_3N_R_origin_wrt_fixedWholebrain.txt
│   │   │       ├── atlasV7_10.0um_scoreVolume_3N_R_4N_R_warp7_DEMO998_detector799_10.0um_scoreVolume_3N_R_4N_R_10.0um_3N_R_surround_200um.bp
│   │   │       ├── atlasV7_10.0um_scoreVolume_3N_R_4N_R_warp7_DEMO998_detector799_10.0um_scoreVolume_3N_R_4N_R_10.0um_3N_R_surround_200um_origin_wrt_fixedWholebrain.txt
│   │   │       ├── atlasV7_10.0um_scoreVolume_3N_R_4N_R_warp7_DEMO998_detector799_10.0um_scoreVolume_3N_R_4N_R_10.0um_4N_R.bp
│   │   │       ├── atlasV7_10.0um_scoreVolume_3N_R_4N_R_warp7_DEMO998_detector799_10.0um_scoreVolume_3N_R_4N_R_10.0um_4N_R_origin_wrt_fixedWholebrain.txt
│   │   │       ├── atlasV7_10.0um_scoreVolume_3N_R_4N_R_warp7_DEMO998_detector799_10.0um_scoreVolume_3N_R_4N_R_10.0um_4N_R_surround_200um.bp
│   │   │       └── atlasV7_10.0um_scoreVolume_3N_R_4N_R_warp7_DEMO998_detector799_10.0um_scoreVolume_3N_R_4N_R_10.0um_4N_R_surround_200um_origin_wrt_fixedWholebrain.txt
│   │   └── atlasV7_10.0um_scoreVolume_warp0_DEMO998_detector799_10.0um_scoreVolume_10.0um
│   │       └── score_volumes
│   │           ├── atlasV7_10.0um_scoreVolume_warp0_DEMO998_detector799_10.0um_scoreVolume_10.0um_12N.bp
│   │           ├── atlasV7_10.0um_scoreVolume_warp0_DEMO998_detector799_10.0um_scoreVolume_10.0um_12N_origin_wrt_fixedWholebrain.txt
│   │           ├── atlasV7_10.0um_scoreVolume_warp0_DEMO998_detector799_10.0um_scoreVolume_10.0um_12N_surround_200um.bp
│   │           ├── atlasV7_10.0um_scoreVolume_warp0_DEMO998_detector799_10.0um_scoreVolume_10.0um_12N_surround_200um_origin_wrt_fixedWholebrain.txt
│   │           ├── atlasV7_10.0um_scoreVolume_warp0_DEMO998_detector799_10.0um_scoreVolume_10.0um_3N_R.bp
│   │           ├── atlasV7_10.0um_scoreVolume_warp0_DEMO998_detector799_10.0um_scoreVolume_10.0um_3N_R_origin_wrt_fixedWholebrain.txt
│   │           ├── atlasV7_10.0um_scoreVolume_warp0_DEMO998_detector799_10.0um_scoreVolume_10.0um_3N_R_surround_200um.bp
│   │           ├── atlasV7_10.0um_scoreVolume_warp0_DEMO998_detector799_10.0um_scoreVolume_10.0um_3N_R_surround_200um_origin_wrt_fixedWholebrain.txt
│   │           ├── atlasV7_10.0um_scoreVolume_warp0_DEMO998_detector799_10.0um_scoreVolume_10.0um_4N_R.bp
│   │           ├── atlasV7_10.0um_scoreVolume_warp0_DEMO998_detector799_10.0um_scoreVolume_10.0um_4N_R_origin_wrt_fixedWholebrain.txt
│   │           ├── atlasV7_10.0um_scoreVolume_warp0_DEMO998_detector799_10.0um_scoreVolume_10.0um_4N_R_surround_200um.bp
│   │           └── atlasV7_10.0um_scoreVolume_warp0_DEMO998_detector799_10.0um_scoreVolume_10.0um_4N_R_surround_200um_origin_wrt_fixedWholebrain.txt
  • Visualize registration. Run python visualize_registration.py NtbNormalizedAdaptiveInvertedGamma example_specs/demo_visualization_per_structure_alignment_spec.json -g example_specs/demo_visualization_global_alignment_spec.json
├── CSHL_registration_visualization
│   └── DEMO998_atlas_aligned_multilevel_down16_all_structures
│       └── NtbNormalizedAdaptiveInvertedGammaJpeg
│           ├── DEMO998_NtbNormalizedAdaptiveInvertedGammaJpeg_225.jpg
│           ├── DEMO998_NtbNormalizedAdaptiveInvertedGammaJpeg_230.jpg
│           └── DEMO998_NtbNormalizedAdaptiveInvertedGammaJpeg_235.jpg

An example registration visualization is given below (white contours = rough global registration, colored = different probability contours of locally registered structures) example registration_visualization