Data

Tools

Indexes

Applications

Experiments

Open Source Science

ventriclescnn

Deep-learning MRI segmentation of ventricles in pediatric hydrocephalus using nnU-Net and VParNet.

https://github.com/0rc0/ventriclescnn

Science Score: 57.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ✓
    CITATION.cff file
    Found CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ✓
    .zenodo.json file
    Found .zenodo.json file
  • ✓
    DOI references
    Found 8 DOI reference(s) in README
  • ○
    Academic publication links
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (7.5%) to scientific vocabulary
Last synced: 6 months ago · JSON representation ·

Repository

Deep-learning MRI segmentation of ventricles in pediatric hydrocephalus using nnU-Net and VParNet.

Basic Info
  • Host: GitHub
  • Owner: 0rC0
  • License: mit
  • Language: Jupyter Notebook
  • Default Branch: main
  • Homepage:
  • Size: 670 KB
Statistics
  • Stars: 0
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Created about 4 years ago · Last pushed 8 months ago
Metadata Files
Readme License Citation

README.md

VentriclesCNN

DOI

OSF.io

License: MIT

Deep-learning MRI segmentation of ventricles in pediatric hydrocephalus using nnU-Net and VParNet.

Kühne, F; Rüther, K; Güttler, C; Stöckel, J; Thomale, U; Tietze, A; Dell’Orco, A "Application of deep neural networks in automatized ventriculometry and segmentation of the aqueduct in pediatric hydrocephalus patients" 2025, OSF Preprints. DOI: [10.17605/OSF.IO/HPU5B](https://doi.org/10.17605/OSF.IO/HPU5B)

Please refer to the README.md in the individual folders

Models' weights are shared on [OSF.io|https://osf.io/hpu5b/]

VentrikelCNN Example Output

Example usage

Create nnunet conda environment

! conda create -f nnunet_conda_env.yaml

Case 1: Segment one single image with one model

``` import glob, os, re from nnunetv2.paths import nnUNetresults, nnUNetraw import torch from nnunetv2.inference.predictfromrawdata import nnUNetPredictor from nnunetv2.imageio.simpleitkreaderwriter import SimpleITKIO models = glob.glob('path-to-models/nUNetTrainernnUNetPlans3dfullres')

predictor = nnUNetPredictor( tilestepsize=0.5, usegaussian=True, usemirroring=True, performeverythingondevice=True, device=torch.device('cuda', 0), verbose=False, verbosepreprocessing=False, allow_tqdm=True )

predictor.initializefromtrainedmodelfolder( models[0], usefolds=(1,), checkpointname='checkpoint_final.pth', )

infile = os.path.join(os.getcwd(),'test/sub-V003ses-01T1w.nii.gz') outfile = os.path.join(os.getcwd(),'test/sub-V003ses-01desc-VentrikelCNN_mask.nii.gz')

predictor.predictfromfiles([[infile]], [outfile], saveprobabilities=False, overwrite=False, numprocessespreprocessing=1, numprocessessegmentationexport=1, folderwithsegsfromprevstage=None, numparts=1, part_id=0) ```

Case 2: Ensemble of all models from five-fold cross-validation

ToDo

Citation

@article{kuhne2025, author = {Kühne, Fabienne and Rüther, Kilian and Güttler, Christopher and Stöckel, Juliane and Thomale, Ulrich-Wilhelm and Tietze, Anna and Dell’Orco, Andrea}, title = {Application of deep neural networks in automatized ventriculometry and segmentation of the aqueduct in pediatric hydrocephalus patients}, year = {2025}, journal = {OSF Preprints}, doi = {10.17605/OSF.IO/HPU5B}, url = {https://doi.org/10.17605/OSF.IO/HPU5B} }

Owner

  • Name: Andrea Dell'Orco
  • Login: 0rC0
  • Kind: user
  • Location: Berlin

Sharing code for neuroimaging research. Credits for profile picture: @lastknight"

Citation (CITATION.cff) 

cff-version: 1.2.0
message: "If you use this software, please cite the following work."
authors:
  - family-names: Kühne
    given-names: Fabienne
    affiliation: "Department of Neonatology, Charité – University Medicine Berlin, Berlin, Germany"
  - family-names: Rüther
    given-names: Kilian
    affiliation: "Institute of Neuroradiology, Charité – University Medicine Berlin, Berlin, Germany"
  - family-names: Güttler
    given-names: Christopher
    affiliation: "Institute of Neuroradiology, Charité – University Medicine Berlin, Berlin, Germany"
  - family-names: Stöckel
    given-names: Juliane
    affiliation: "Institute of Neuroradiology, Charité – University Medicine Berlin, Berlin, Germany"
  - family-names: Thomale
    given-names: Ulrich-Wilhelm
    affiliation: "Department of Neurosurgery with Pediatric Neurosurgery, Charité – University Medicine Berlin, Berlin, Germany"
  - family-names: Tietze
    given-names: Anna
    affiliation: "Institute of Neuroradiology, Charité – University Medicine Berlin, Berlin, Germany"
  - family-names: Dell’Orco
    given-names: Andrea
    affiliation: "Institute of Neuroradiology, Charité – University Medicine Berlin, Berlin, Germany"
title: "Application of deep neural networks in automatized ventriculometry and segmentation of the aqueduct in pediatric hydrocephalus patients"
version: "1.0.0"
date-released: 2025-03-28
doi: "10.17605/OSF.IO/HPU5B"
url: "https://osf.io/hpu5b/"

GitHub Events

Total
  • Push event: 5
  • Public event: 1
  • Create event: 1
Last Year
  • Push event: 5
  • Public event: 1
  • Create event: 1