PyDactim Tutorial

Here is some example of what you can do with PyDactim 🔥

🔧 Pipeline example

Let’s imagine a small pipeline, including:

  • Auto cropping the MRI image

  • Normalizing voxel intensity value by 1

  • Extracting the brain

  • Applying a SUSAN filter

So first at all, let’s import PyDactim and define a variable of some kind of data:

import pydactim as pyd

data_path = "path/to/your/data.nii.gz"

Now we can call the auto crop function:

cropped_data_path, cropped_idx = pyd.crop(data_path)

As you can see, all PyDactim function take a path as input and output the generated image as a path. The new image is saved in the same location than the input and will have a suffix added that you can change according to the parameter suffix. For instance, let’s change the suffix for the crop:

cropped_data_path, cropped_idx = pyd.crop(data_path, suffix='zoom')

That’s a bad suffix ! But now, in the same folder than the input image you will have the generated image with this suffix.

If you want to overwrite an already existing output image, you can use the parameter force.

cropped_data_path, cropped_idx = pyd.crop(data_path, force=True)

Let’s see now the rest of the pipeline:

cropped_data_path, cropped_idx = pyd.crop(data_path) # This will crop the data to remove most of the noise
normalized_data_path = pyd.normalize(cropped_data_path) # Normalize between 0 and 1 the image
brain_data_path, mask_data_path = pyd.skull_stripping(normalized_data_path, model_path="path_to_fold", mask=True) # See https://github.com/MIC-DKFZ/HD-BET for the fold of the HD-BET
susan_data_path = pyd.susan(brain_data_path, offset=3) # SUSAN filtering with a kernel of size 3

💻 Viewer App

PyDactim also have a GUI to visualize and launch the function without using coding. It can be easily launch with the following line:

app = ViewerApp()

It can handle 3D MRI images as well as 4D Perfusion MRI data with a ROI to observe the signal over time. This GUI is still in work so you might encounter some unexpected behaviour.

📁 Managing DICOM files

You can also sort DICOM by tags such as Study description, Patient ID, Study Date, Series description:

dicom_dir = "directory/to/your/dicom/"
sorted_dir = "directory/for/the/sorted/dicom/"

sort_dicom(dicom_dir, sorted_dir)

Then, you can convert a directory of DICOM in NIfTI

dicom_dir = "directory/to/your/dicom/"
nifti_dir = "directory/for/the/sorted/dicom/"

convert_dicom_to_nifti(dicom_dir, nifti_dir)