exodeepfinder 0.2.2

ExoDeepFinder is an original deep learning approach to localize macromolecules in cryo electron tomography images. The method is based on image segmentation using a 3D convolutional neural network.

ExoDeepFinder

ExoDeepFinder is an exocytosis event detection tool.

This work is based on DeepFinder, customized for use in TIRF microscopy.

Installation guide

It is strongly advised to create a virtual environment for ExoDeepFinder before installing it. With virtualenv, simply run python -m venv exoDeepFinder/ to create your environment, and source exoDeepFinder/bin/activate to activate it.

Then, you can install DeepFinder with pip:

pip install exodeepfinder

Also, in order for Tensorflow to work with your Nvidia GPU, you need to install CUDA. An alternative could be to install the python packages cudatoolkit and cudnn. Once these steps have been achieved, the user should be able to run DeepFinder.

Usage

To detect exocytose events, you can either use the pretrained model to generate segmetations, or you can train your own model from your annotated images.

For more information about the different ExoDeepFinder commands, use the --help option. For example, run edf_detect_spots --help to get more information about the edf_detect_spots command.

Exocytose events segmentation

ExoDeepFinder handles movies made from tiff files, where each tiff file is a frame of the movie, and their name ends with the frame number ; like in the following structure:

exocytose_data/
├── movie1/
│   ├── frame_1.tiff
│   ├── frame_2.tiff
│   └── ...

The frame extensions can be .tif, .tiff, .TIF or .TIFF.

The movie folders (containing the frames in tiff format) can be converted into a single .h5 file with the edf_convert_tiff_to_h5 command. Most ExoDeepFinder commands take h5 files as input, so the first step is to convert the data to h5 format with the following command: edf_convert_tiff_to_h5 --tiff path/to/movie/folder/ --output path/to/output/movie.h5

You can also generate all your movie folders at once using the --batch option:

edf_convert_tiff_to_h5 --batch path/to/movies/ --output path/to/outputs/ --make_subfolder

where path/to/movies/ contains movies folders (which in turn contains tiff files). The --make_subfolder option enable to put all tiff files in a till/ subfolder ; which is useful in batch mode. The --batch option enables to process multiple movie folders at once and work in the same way in all ExoDeepFinder commands.

The above command will turn the following file structure:

exocytose_data/
├── movie1/
│   ├── frame_1.tiff
│   ├── frame_2.tiff
│   └── ...
├── movie2/
│   ├── frame_1.tiff
│   └── ...
└── ...

into this one:

exocytose_data/
├── movie1/
│   ├── tiff/
│   │   ├── frame_1.tiff
│   │   ├── frame_2.tiff
│   │   └── ...
│   └── movie.h5
├── movie2/
│   ├── tiff/
│   |   ├── frame_1.tiff
│   │   └── ...
│   └── movie.h5
└── ...

To generate segmentations, you can either use the napari-exodeepfinder plugin which provide a simple graphical interface, or you can run the following command lines.

To segment a movie, use: edf_segment --movie path/to/movie.h5 --model_weights examples/analyze/in/net_weights_FINAL.h5 --patch_size 160 --visualization

This will generate a segmentation named path/to/movie_semgmentation.h5 with the pretrained weigths in examples/analyze/in/net_weights_FINAL.h5 and patches of size 160. It will also generate visualization images.

See edf_segment --help for more information about the input arguments.

To cluster a segmentation and create an annotation file from it, use: edf_generate_annotation --segmentation path/to/movie_segmentation.h5 --cluster_radius 5

Using the GUI

The napari-deepfinder plugin can be used to perform perictions. Open the movie you want to segment in napari (it must be in h5 format). In the menu, choose Plugins > Napari DeepFinder > Segmentation to open the segmentation tools. Choose the image layer you want to segment. Select the examples/analyze/in/net_weights_FINAL.h5 net weights ; or the path of the model weights you want to use for the segmentation. Use 3 for the number of class, and 160 for the patch size. Choose an output image name (with the .h5 extension), then launch the segmentation.

Training

To train a model, your data should be organized in the following way:

exocytose_data/
├── movie1/
│   ├── frame_1.tiff
│   ├── frame_2.tiff
│   └── ...
├── movie2/
│   ├── frame_1.tiff
│   └── ...
└── ...

There is no constraint on the file names, but they must contain the frame number (the last number in the file name must be the frame number), and be in the tiff format (it could work with other format like .png since images are read with the skimage.io.imread() function of the scikit-image library). For example frame_1.tiff could also be named IMAGE32_1.TIF. Similarly, there is no constraint on the movie names.

For each movie, tiff files must be converted to a single .h5 using the edf_convert_tiff_to_h5 command:

edf_convert_tiff_to_h5 --batch path/to/exocytose_data/ --make_subfolder

This will change the exocytose_data structrue into the following one:

exocytose_data/
├── movie1/
│   ├── tiff/
│   │   ├── frame_1.tiff
│   │   ├── frame_2.tiff
│   │   └── ...
│   └── movie.h5
├── movie2/
│   ├── tiff/
│   |   ├── frame_1.tiff
│   │   └── ...
│   └── movie.h5
└── ...

Then, bright spots must be detected in each frame with a spot detector such as Atlas (it can be another spot detector). The Atlas installation instructions are detailed in the repository.

Once your installed atlas, you can detect spots in each frame using the edf_detect_spots command:

edf_detect_spots --detector_path path/to/atlas/ --batch path/to/exocytose_data/

This will generate detector_segmentation.h5 files (the semgentations of spots) in the movie folders:

exocytose_data/
├── movie1/
│   ├── tiff/
│   │   ├── frame_1.tiff
│   │   ├── frame_2.tiff
│   │   └── ...
│   ├── detector_segmentation.h5
│   └── movie.h5
├── movie2/
└── ...

You can make sure that the detector segmentations are correct by opening them in napari with the corresponding movie. Open both .h5 files in napari, put the detector_segmentation.h5 layer on top, then right-click on it and select "Convert to labels". You should see the detections in red on top of the movie.

Annotate the exocytose events in the movies with the napari-deepfinder plugin. Follow the install instructions, and open napari. In the menu, choose Plugins > Napari DeepFinder > Annotation to open the annotation tools. Open a movie (for example exocytose_data/movie1/movie.h5). Create a new points layer, and name it movie_1 (any name with the _1 suffix, since we want to annotate with the class 1). In the annotation panel, select the layer you just created in the "Points layer" select box. You can use the Orthoslice view to easily navigate in the volume, by using the Plugins > Napari DeepFinder > Orthoslice view menu. Scroll in the movie until you find and exocytose event. If you opened the Orthoslice view, you can click on an exocytose event to put the red cursor at its location, then click the "Add point" button in the annotation panel to annotate the event. You can also use the "Add points" and "Delete selected point" button from the layer controls. When you annotated all events, save your annotations to xml by choosing the File > Save selected layer(s)... menu, or by using ctrl+S (command+S on a mac), and choose the Napadi DeepFinder (*.xml) format. Save the file beside the movie, and name it expert_annotation.xml (this should result in the exocytose_data/movie1/expert_annotation.xml with the above example). Annotate all training and validation movies with this procedure ; you should end up with the following folder structure:

exocytose_data
├── movie1/
│   ├── tiff/
│   │   ├── frame_1.tiff
│   │   ├── frame_2.tiff
│   │   └── ...
│   ├── detector_segmentation.h5
│   ├── expert_annotation.xml
│   └── movie.h5
├── movie2/
└── ...

Make sure that the expert_annotation.xml files you just created have the following format:

<objlist>
  <object tomo_idx="0" class_label="1" x="71" y="152" z="470"/>
  <object tomo_idx="0" class_label="1" x="76" y="184" z="445"/>
  <object tomo_idx="0" class_label="1" x="141" y="150" z="400"/>
  <object tomo_idx="0" class_label="1" x="200" y="237" z="420"/>
  <object tomo_idx="0" class_label="1" x="95" y="229" z="438"/>
  ...
</objlist>

The class_label must be 1, and tomo_idx must be 0.

Use the edf_generate_segmentation command to convert the annotations to segmentations:

edf_generate_segmentation --batch path/to/exocytose_data/

You will end up with the following structure:

exocytose_data/
├── movie1/
│   ├── tiff/
│   │   ├── frame_1.tiff
│   │   ├── frame_2.tiff
│   │   └── ...
│   ├── detector_segmentation.h5
│   ├── expert_annotation.xml
│   ├── expert_segmentation.h5
│   └── movie.h5
├── movie2/
└── ...

Again, you can check on napari that everything went right by opening all images and checking that expert_segmentation.h5 corresponds to expert_annotation.xml and the movie.

Then, merge detector detections with expert annotations with the edf_merge_detector_expert command:

edf_merge_detector_expert --batch path/to/exocytose_data/

This will create two new files merged_annotation.xml (the merged annotations) and merged_segmentation.h5 (the merged segmentations). The exocytose events are first removed from the detector segmentation (detector_segmentation.h5), then the remaining events (from the dector and the expert) are transfered to the merged segmentation (merged_segmentation.h5), with class 2 for exocytose events and class 1 for others events. The maximum number of other events in the annotation is 9800 ; meaning that if there are more than 9800 other events, only 9800 events will be picked randomly and the others will be discarded.

The exocytose_data/ folder will then follow this structure:

exocytose_data/
├── movie1/
│   ├── tiff/
│   │   ├── frame_1.tiff
│   │   ├── frame_2.tiff
│   │   └── ...
│   ├── detector_segmentation.h5
│   ├── expert_annotation.xml
│   ├── expert_segmentation.h5
│   ├── merged_annotation.xml
│   ├── merged_segmentation.h5
│   └── movie.h5
├── movie2/
└── ...

Again, make sure everything looks right in napari.

Finally, the training data should be organized in the following way:

dataset/
├── train/
│   ├── movie1.h5
│   ├── movie1_objl.xml
│   ├── movie1_target.h5
│   ├── movie2.h5
│   ├── movie2_objl.xml
│   ├── movie2_target.h5
...
└── valid/
    ├── movie3.h5
    ├── movie3_objl.xml
    ├── movie3_target.h5
...

Use edf_structure_training_dataset to convert the current folder structure into the training structure:

edf_structure_training_dataset --input path/to/exocytose_data/ --output path/to/dataset/

This will organize the files with the above structure, by putting 70% of the movies in the train/ folder, and 30% of them in the valid/ folder.

Make sure the output folder is correct, and that you can open its content in napari.

Finally, launch the training with edf_train --dataset path/to/dataset/ --output path/to/model/.

To sum up, here is all the steps you should execute to train a new model:

1. Convert tiff frames to h5 file: edf_convert_tiff_to_h5 --batch path/to/exocytose_data/ --make_subfolder
2. Use napari-exodeepfinder to annotation exocytose events in the movies
3. Detect all spots: edf_detect_spots --detector_path path/to/atlas/ --batch path/to/exocytose_data/
4. Generate detector segmentations: edf_generate_segmentation --batch path/to/exocytose_data/
5. Merge expert and detector segmentation: edf_merge_detector_expert --batch path/to/exocytose_data/
6. Structure the files: edf_structure_training_dataset --dataset path/to/exocytose_data/ --training path/to/dataset/
7. Train the model: edf_train --dataset path/to/dataset/ --output path/to/model/