Prostate-Needle-Finder

Prostate-Needle-Finder

‹ Go back to models

Model parameters

  • Task: Biopsy needle trajectory and tip detection in intraoperative MRIs.
  • Imaging modality: MRI
  • Organ: Prostate
  • Input(s): 1) Pelvic T2-W MRI, 2)Prostate gland segmentation (Approximate)
  • Output(s): 1)Biopsy Needle Segmentation, 2)Fiducial list containing needle tip
  • Training set size: 50 patients (410 MRIs) / Test set size: 21 patients (173 MRIs)
  • Performance: Mean needle tip localization error: 2.8 mm
  • contact: Alireza Mehrtash, email: mehrtash at bwh.harvard.edu
  • Size: 6.8 GB

Scientific Publication

Details for Prostate-Needle-Finder model are provided in the following scientific publication:

Article under review.

BibTeX citation

@inproceedings{,
  title={},
  author={},
  booktitle={},
  volume={},
  year={},
  organization={}
}

License

Prostate-Needle-Finder model is licensed under Slicer License.

Demo

Command-line interface guide

Download Docker Image

docker pull deepinfer/prostate

Example

docker run -t -v ~/data/needle_test/:/data deepinfer/prostate\
                   --ModelName prostate-needle-finder\
                   --InputVolume /data/confirmation_volume.nrrd\
                   --InputProstateMask /data/prostate_label.nrrd\
                   --OutputLabel /data/needle_label.nrrd \
                   --OutputFiducialList /data/Tip.fcsv\
                   --InferenceType Single\
                   --verbose

Inputs

[Mandatory]
ModelName: ('prostate-needle-finder')
InputVolume: (an existing filename locating the T2-Weighted Pelvic MRI containing Needle)
InputProstateMask: (an existing filename locating the rough prostate gland location)
OutputLabel: (output path of the needle label)
OutputFiducialList: (output path of the fiducial list in fcsv format (slicer fiducial list format) where the needle
tip will be saved)
Inference: (Single, Ensemble)
    Single: the prediction would be the output of a single model. Ensemble: the prediction will be the 
    calculated by ensembling of 5 models from 5-fold cross validation and majority voting.

[Optional]
verbose : 
verbose mode for printing additional details about the procedure.