BIIG-UC3M / Docker_image_pipeline

Docker for image pipeline

Latest version 1.1

How to use it:

docker run -it -v /your/path/to/Images:/app/Images image_pipeline/call_apply -v /your/path/to/Results:/app/Results [image] [species*]

At this moment, this docker image returns three files for each CT image in the Images folder:

• First of all, two of the three files are the resulting mask in .mhd and .raw format of segmenting the lungs 🫁 (left and right lobes). MHD file is an ITK MetaImage Header. Insight Segmentation and Registration Toolkit (ITK).

• Last but not least, a comma-separated values file (.csv) with the value of volume of healthy and unhealthy lung.

Input images can be in .mhd/.raw or DICOM format.

It is also advisable to run the docker typing which model (species) is going to be used to get the masks. Otherwise, default model is human's model. There are two additional models: mice and macaques.

---

---

Packaging for DDIM

Usage:

python3 packaging_for_DDIM.py <path_to_images_folder>

When uploading an experiment to DDIM, it is necessary to upload a .zip file with with the following structure:

Furthermore, it is mandatory to create a .csv or .xls (Excel) file with some metadata about the images.

Job_ID	Molecule	Technique	Experiment_ID	Institute	Owner	Date	Time	Subject_ID	Linked_Experiment_ID

To ease this process, packaging_for_DDIM.py can be used.

In order to do so, in <path_to_images_folder> there must be one folder for each subject and its images inside that.

For example:

DICOM Tags

A DICOM data element, or attribute, is composed of a tag that identifies the attribute, usually in the format (XXXX,XXXX). To provide all necessary fields to create metadata file, the following DICOM tags are used:

    (0040,0280)  # User-defined comments on the Performed Procedure Step. Used for Job_ID
    (0010,1000)  # Other Patient IDs. Used for Molecule
    (0008,0060)  # Modality
    (0020,0010)  # Study ID, for human consumption
    (0008,0090)  # Referring Physician's Name. Used for Institute
    (0008,0020)  # Date in format YYYYMMDD
    (0008,0030)  # Time in format HHMMSS
    (0010,0020)  # Patient ID
    (0010,4000)  # User-defined additional information about the Patient. Used for TB strain.

If images have got these tags, packaging_for_DDIM.py will use them (asked previously if user wants to use these fields). If one of them is missing, it will be asked to the user to type them.

As a result, the .zip file will be produced and ready to upload it to DDIM.