nic-vicorob - WMH Segmentation Challenge
Skip to content

WMH Segmentation Challenge

Grand Challenge

  • Home
  • Details
  • Data
    • Manual reference standard
    • MRI parameters
  • Methods
    • Submit
    • Dockerfile
    • Example: python
    • Example: matlab
  • Results
    • Evaluation
    • Teams:
    • achilles
    • acunet
    • arg
    • bigrbrain
    • bigrbrain 2
    • bioengineering_espol_team
    • buckeye_ai
    • caai_amh
    • cian
    • coroflo
    • dice
    • fmrib-truenet
    • fmrib-truenet 2
    • hadi
    • himinn
    • ipmi-bern
    • k2
    • knight
    • livia
    • lrde
    • misp
    • misp 2
    • neuro.ml
    • neuro.ml 2
    • nic-vicorob
    • nih_cidi
    • nih_cidi 2
    • nist
    • nlp_logix
    • pitt-vamp
    • pitt-vamp 2
    • pgs
    • rasha_simple
    • rasha_improved
    • scan
    • skkumedneuro
    • sysu_media
    • sysu_media 2
    • text_class
    • tig
    • tig 2
    • tignet
    • tum_ibbm
    • uned
    • uned 2
    • uned_contrast
    • upc_dlmi
    • wta
    • wta 2
  • Organizers
  • —
  • Login
  • Register

nic-vicorob

A 10-layer 3D convolutional neural network architecture previously used to segment multiple sclerosis lesions. A cascaded training procedure was employed, training two separate networks to first identify candidate lesion voxels and next to reduce false positive detections. A third network re-trains the last fully connected layer to perform WMH segmentation.

Presentation of this method.
Full description of this method.
Docker container of this method.

by Hugo J. Kuijf, Image Sciences Institute, UMC Utrecht, the Netherlands

Privacy Policy