Use cases - raidionics/Raidionics GitHub Wiki

Version 1.1

1. Single patient

To process a handful of patients, either to generate the tumor segmentation or the complete clinical report, and interact with the results.

  • Make sure the proper MRI sequence type is filled in correctly for each MRI series in the right-hand panel. The FLAIR MRI scan will be used when processing a diffuse lower-grade glioma while the contrast-enhanced T1-weighted MRI scan (T1-CE) will be used for the other tumor types.
    ⚠️ If multiple MRI scans for the same sequence type are loaded, the first one will be used automatically to compute the results.
  • During processing, the right-hand panel shows the progress and the remaining amount of steps yet to perform.
  • Upon process completion, all results are stored in the patient folder automatically (indicated by the folder location) and are populated inside the right-hand panel.
    ❗ By default the central GUI is reset upon process completion, so the automatic segmentations/results generated are not directly displayed. You must navigate the Annotations and Structures sections of the right-hand panel and toggle the masks you are willing to view (only the results associated with the displayed MRI Series are listed).

2. Study/batch mode

To load a collection/cohort of patients (e.g., as part of a study), and process all of them sequentially.

2.1. Data loading

Several possibilities are currently available, linked to the following four buttons available in the GUI: Single, Cohort, Single DICOM, Cohort DICOM. In the DICOM cases, all possible MRI scans will be extracted from the folder.
⚠️ The user will then be required to manually set the proper MRI sequence tags for each loaded MRI scan, or remove unwanted scans.

  • Single: the selected folder location contains multiple volume files (e.g., nifti, nrrd) corresponding to the same patient.
  • Single DICOM: the selected folder location points to a root DICOM folder for a singular patient. The folder can contain only the DICOM files corresponding to a single MRI volume, but also multiple investigations.
  • Cohort: the selected folder location contains: (i) a collection of sub-folders where each sub-folder corresponds to a singular patient and can contain multiple volume files, or (ii) a collection of volume files whereby each corresponds to a singular patient.
  • Cohort DICOM: the selected folder location points to a folder containing multiple sub-folders, each sub-folder corresponding to a DICOM folder for a singular patient.

Version 1.0

Usage

Two modes are proposed: (i) Single-use for processing one MRI scan at a time with the possibility to view and interact with the results, and (ii) Batch-mode for processing multiple MRI scans in a row, without any vizualisation.

For the single use case:

  1. Click 'Input MRI...' to select from your file explorer the MRI scan to process (unique file), preferably as nifti (*.nii, .nii.gz).
    1
    ) Alternatively, Click 'File > Import DICOM...' if you wish to process an MRI scan as a DICOM sequence.
  2. Click 'Output destination' to choose a directory where to save the results.
  3. (OPTIONAL) Click 'Input segmentation' to choose a tumor segmentation mask file, if nothing is provided the internal model with generate the segmentation automatically.
  4. Select the tumor type from the drop-down menu, supported types are: (i) High-Grade Glioma (glioblastoma), (ii) Low-Grade Glioma, (iii) Meningioma, and (iv) Metastasis.
  5. Click 'Run segmentation' to generate the brain tumor mask, or 'Run standard reporting' to perform the full analysis. The human-readable version of the results will be displayed directly in the interface.

Generated results

The output folder is populated automatically with the following:
* The diagnosis results in human-readable text (report.txt) and Excel-ready format (report.csv).
* The automatic segmentation masks of the brain and the tumor in the original patient space (input_brain_mask.nii.gz and input_tumor_mask.nii.gz).
* The cortical structures mask in original patient space for the different atlases used.
* The input volume and tumor segmentation mask in MNI space in the sub-directory named 'registration'.

⚠️ **GitHub.com Fallback** ⚠️