Difference between revisions of "2011 Winter Project Week: Python and Slicer4"

From NAMIC Wiki
Jump to: navigation, search
 
(4 intermediate revisions by 2 users not shown)
Line 55: Line 55:
  
 
{|
 
{|
|[[image:LabelStatistics-Module-2011-01-09.png|thumb|500px|Label Statistics running as a slicer module]].
+
|[[image:LabelStatistics-Slicelet-2011-01-09.png|thumb|Label Statistics running stand-alone as a Slicelet]].
|[[image:LabelStatistics-Slicelet-2011-01-09.png|thumb]500px|Label Statistics running stand-alone as a Slicelet]].
+
|[[image:LabelStatistics-Module-2011-01-09.png|thumb|Label Statistics running as a slicer module]].
 +
|[[image:DWI2Tractography-Slicelet-2011-01-14.png|thumb|DWI to full brain tractography step 1]].
 +
|[[image:DWI2Tractography2-Slicelet-2011-01-14.png|thumb|DWI to full brain tractography step 2]].
 +
|[[image:DWI2Tractography3-Slicelet-2011-01-14.png|thumb|DWI to full brain tractography step 3]].
 
|}
 
|}
 +
  
 
Command line to run "Slicelet" from Slicer4 build directory:  
 
Command line to run "Slicelet" from Slicer4 build directory:  
  
 
<code>
 
<code>
Slicer4 lib/Slicer3/QTScriptedModules/LabelStatistics.py
+
% Slicer4 lib/Slicer3/QTScriptedModules/LabelStatistics.py
 
</code>
 
</code>
 +
 +
* TODO Items:
 +
** pass command line options to slicelets (use tclap-style --xml like the CLI modules)
 +
** optimize startup time for the interpreter
 +
** expose remaining parts of the Qt/qSlicer API to Python
 +
** progress and status output
 +
** wrapper to run non-scripted modules in slicelet mode
 +
 +
* Architectural Issues
 +
** Object inheritance plan for scripted modules and slicelets
 +
** Utilities needed to make it easy for developers/users to put together pipelines that mix module types
 +
  
 
</div>
 
</div>

Latest revision as of 17:34, 14 January 2011

Home < 2011 Winter Project Week: Python and Slicer4

Key Investigators

  • Jc, Danielle - Kitware
  • Jim - GE
  • Steve - Isomics
  • Demian - BWH

Objective

Develop scripting patterns for implementing workflows in slicer4 to port slicer3 functionality and streamline implementation of new functionality.

Approach, Plan

  • Review initial experiences with PythonQt in slicer4
    • Editor and SWidgets
    • TubeTK-based Interactive PDF Segmenter
    • Endoscopy
  • Consider general architecture for creating module GUIs (scripted widgets)
    • allow adaptive re-use of widgets in different modules such as:
      • embed subset of painting tools inside interactive segmenters
      • embed small model maker interface in segmentation tools
      • embed some annotation (e.g. fiducial) for landmarking in registration
      • embed CLI interfaces in workflow wizards
    • define inheritance for PythonQt implementations to provide consistency across module GUI implementaitons
  • Consider use cases for (scripted) slicer modules - can one implementation pattern cover all use cases?
    • embed GUI in slicer interface (parent is slicer module panel)
    • run as part of a workflow (parent is one step of workflow)
    • run GUI stand-alone (module's widget is top level window)- call this Slicelet mode.
    • run without GUI (parameters passed on command line CLI-style)

Progress

Slicer4 developer beta supports PythonQt scripted modules.

References

Example LabelStatistics module source code.

Label Statistics running stand-alone as a Slicelet
.
Label Statistics running as a slicer module
.
DWI to full brain tractography step 1
.
DWI to full brain tractography step 2
.
DWI to full brain tractography step 3
.


Command line to run "Slicelet" from Slicer4 build directory:

% Slicer4 lib/Slicer3/QTScriptedModules/LabelStatistics.py

  • TODO Items:
    • pass command line options to slicelets (use tclap-style --xml like the CLI modules)
    • optimize startup time for the interpreter
    • expose remaining parts of the Qt/qSlicer API to Python
    • progress and status output
    • wrapper to run non-scripted modules in slicelet mode
  • Architectural Issues
    • Object inheritance plan for scripted modules and slicelets
    • Utilities needed to make it easy for developers/users to put together pipelines that mix module types