Skip to content

Latest commit

 

History

History
40 lines (32 loc) · 3 KB

README.md

File metadata and controls

40 lines (32 loc) · 3 KB
Raw

Step 3: Hierarchically arterial and cortical volume feature extration

We extract arterial and cortical features of every subject in terms of 3 hierarchical levels: 1) whole brain, 2) four common vascular regions including anterior cortical region (ACR), posterior cortical region (PCR), middle cortical region (MCR), and circle of willis region (CoWR), and 3) conventional 82 Brodmann areas.

System requirements

Operating System

Windows 10.

Hardware requirements

CPU with 4 cores and 16GB RAM are enough.

Installation instructions

  1. Install Matlab and third party packages SPM12, DPABI and NIFTI following the instructions in Step 1, if you haven't installed them.
  2. Download this 3_feature_extraction folder under the same directory as 1_preprocessing folder to your device, and also add utils folder and spmutils folder to the default search path of Matlab.

How to run it?

  1. Copy the predicted MRA segmentations from the Pred_folder in Step2 to current Windows device.

  2. Coregister raw MRA volumes (along with their prediction) to individual T1 space.

    • set the variable predfile_dir to the directory for saving predicted segmentations in main_coreg_MRA_to_T1.m.
    • run main_coreg_MRA_to_T1.m, the resulted every paired coregisterred MRA volume and segmentation will be saved in the corresponding ./1_preprocessing/MRA/subjectXXXX folder. And all resulted coregisterred MRA volumes will be further copied to a subfolder ./coregMRAs.
    • copy all coregisterred MRA segmentations to a folder.

  3. Run segmentation inference of all coregisterred MRA volumes following Step 2.

    • copy the above resulted coregisterred MRA volumes to a computer or server with linux system.
    • run command line: 
        CUDA_VISIBLE_DEVICES=0 python predict_simple.py -i MRA_folder -o Pred_folder2 -t 503 -tr nnUNetTrainerSSLTune -m 3d_fullres -f 0 -p nnUNetPlansv2.1
      Here MRA_folder should be the absolute path of coregisterred MRA volumes. Pred_folder can be defined according to the user's willingness. Segmentation inference of all coregisterred MRA volumes will be saved in the Pred_folder.
    • copy segmentation inference back.

  4. Combine coregisterred MRA segmentation and segmentation for coregisterred MRA volume:

    • set the variable pred_of_coregMRA_dir to the folder containing segmentation results of all coregisterred MRA volumes in 3.
    • set the variable coreg_of_predMRA_dir to the folder containing all coregisterd MRA segmentation in 2.
    • run main_combine_vascular_segmentations.m to combine the two segmentation results for all subjects.

  5. Extract hierarchical cortical and vascular volumes from predefined atlas.

    • run main_extract_features.m. Extracted features will be saved in a .cvs file for further statistical analysis.