Difference between revisions of "EMSegmenter-Tasks"

From Slicer Wiki
Jump to: navigation, search
Line 101: Line 101:
 
WM: white matter <br>
 
WM: white matter <br>
 
CSF: cerebrospinal fluid <br>
 
CSF: cerebrospinal fluid <br>
cau: caudate
+
cau: caudate <br>
put: putamen
+
put: putamen <br>
hip: hippocampus
+
hip: hippocampus <br>
ICC: Intracranial Cavity
+
ICC: Intracranial Cavity <br>
 
WGC: <br>
 
WGC: <br>
 
lSGM: <br>
 
lSGM: <br>

Revision as of 22:38, 23 November 2010

Home < EMSegmenter-Tasks

Under construction

What is a task?

Most clinicians are using the EMSegmenter to segment specific parts of the human body. Those segmentations depend heavily on the used input data. The signal level in the input data varies with the used image modalities (e.g. MRI, CT, ...) and with each anatomical structure. The EMSegmenter can take advantage of some knowledge about the signal level of those anatomical structures in the different image modalities.
Technically, each tasks consist of a .mrml file and a .tcl file.
The .mrml file stores the anatomical properties (mean values, covariance values) in a user defined tree structure.
The .tcl script is used to perform some pre-processing on the input data.


Existing Tasks

There are currently 2 tasks defined in Slicer 3.6.2

MRI Human Brain

Description

MRI Human Head pipeline for a finer-grained parcellation based on data provided by

Anatomical Tree

  • ROOT
    • BG
      • AIR
      • Skull
    • ICC
      • WM
      • GM
      • CSF

Result

MRI-Human-Brain BeforeAfter.png


MRI Human Brain Parcellation

Description

MRI Human Head pipeline for a finer-grained parcellation based on data provided by

Anatomical Tree

Legend:
BG: background
LT: left temporal
GM: grey matter
WM: white matter
CSF: cerebrospinal fluid

Tree:

  • Root
    • BG
    • GM
      • LTGM
        • LTGM1
        • LTGM2
        • LTGM3
        • LTGM4
      • RTGM
        • RTGM1
        • RTGM2
        • RTGM3
        • RTGM4
      • SUBGM
    • WM
      • LTWM
        • LTWM1
        • LTWM2
        • LTWM3
        • LTWM4
      • RTWM
        • RTWM1
        • RTWM2
        • RTWM3
        • RTWM4
      • SUBWM
    • CSF

Result

MRIHumanBrainParcellation.png

Collaborators

Padmapriya Srinivasan and Sylvain Bouix (PNL-BWH)


Tasks in development

Non-Human Primate

Legend

BG: background
LT: left temporal

GM: grey matter
WM: white matter
CSF: cerebrospinal fluid
cau: caudate
put: putamen
hip: hippocampus
ICC: Intracranial Cavity
WGC:
lSGM:
rSGM:
BS:
CB:

Anatomical Tree

  • Root
    • Head
      • ICC
        • WGC
          • GM
            • lSGM
              • lcau
              • lput
              • lhip
            • rSGM
              • rcau
              • rput
              • rhip
            • GMleaf
          • WM
          • CSF
        • BS
        • CB
      • rest

Result

NHP.png


Collaborators

Andriy Fedorov (BWH)

CT Hand Bone

Collaborators

TBD

Task creation process - for developer

The goal is to create a task with the name 'Hello World'

After following the instructions below the new task will consist of following files:

./Slicer3/Modules/EMSegment/Tasks/Hello-World.mrml
./Slicer3/Modules/EMSegment/Tasks/Hello-World.tcl
./Slicer3/Modules/EMSegment/Tasks/Hello-World/<atlas_file_1>
./Slicer3/Modules/EMSegment/Tasks/Hello-World/<atlas_file_2>
[...]
./Slicer3/Modules/EMSegment/Tasks/Hello-World/<atlas_file_i>
[...]
./Slicer3/Modules/EMSegment/Tasks/Hello-World/<atlas_file_N>


  • Create a new directory ./Slicer3/Modules/EMSegment/Tasks/Hello-World
  • Copy the atlas files for air, GM, T1 into this directory.
  • Start Slicer
  • Switch to the EMSegmenter module
  • Create new task
    • Name: Hello World
    • Pre-processing: None
  • Basically follow our EMSegmenter advanced tutorial, but create a simple tree
    • load ./Slicer3/Modules/EMSegment/Testing/TestData/MiscVolumeData/MRIHumanBrain_T1_aligned.nrrd
    • Add a channel, call it T1, assign MRIHumanBrain_T1_aligned
    • Create a simple anatomical tree (air and GM)
    • Load atlas files only from ./Slicer3/Modules/EMSegment/Tasks/Hello-World/
    • generate mean and covariance values
    • specify node weights
  • At step 9/9. click on 'Create Template File'
    • Save the file under ./Slicer3/Modules/EMSegment/Tasks/Hello-World.mrml
    • The target node and some other unneeded nodes will be removed from the mrml scene and written to disk.
  • Close Slicer
  • Copy our standard tcl file ./Slicer3/Modules/EMSegment/Tasks/Template.tcl to ./Slicer3/Modules/EMSegment/Tasks/Hello-World.tcl
  • Adjust this tcl file to include your pre-processing pipeline.
  • edit ./Slicer3/Modules/EMSegment/Tasks/Hello-World.tcl to specify your own pre-processing
  • edit ./Slicer3/Modules/EMSegment/Tasks/Hello-World.mrml to use your own pre-processing
    • - <EMS TclTaskFilename="GenericTask.tcl" ></EMS>
    • + <EMS TclTaskFilename="Hello-World.tcl" ></EMS>
  • touch ../Slicer3/Modules/EMSegment/CMakeLists.txt and do a 'make' to copy the new files into the binary directory