SimITK and SimVTK: ITK and VTK in Simulink

1
SimITK and SimVTKITK and VTK in Simulink

• DG Gobbi, P Mousavi, KM Li, J Xiang, A
  Campigotto,
• A LaPointe, G Fichtinger, P Abolmaesumi
• Medical Image Analysis Laboratory (MedIA)
• http//www.media.queensu.ca
• School of Computing
• Queens University

2
Motivation

• Visual programming for ITK and VTK
• Drag drop approach to application development
• Increase understandability maintainability
• Instructional value
• MATLAB is our friend
• Use MATLABs Simulink visual environment
• Hence, avoid reinventing the wheel
• Allow mix of MATLAB code and ITK / VTK

3
Related Work

• Visual Programming of VTK and ITK
• ITKBoard (CSIRO)
• SciRun (Utah)
• MeVisLab (MeVis Research GmbH)
• caBIG XIP (caBIG, Siemens, Washington U. in St
  Louis)
• ITK and VTK in MATLAB (without Simulink)
• MATITK (Chu and Hamarneh)
• VTK plus MATLAB various
• Us Visual ITK/VTK Programming in MATLAB

4
MATLABs Simulink Environment
Library
Model
5
Our Own Simulink Libraries
6
3D Processing with ITK
7
Combine ITK Processing and VTK Rendering
Output VTK Window
Input cthead1.png
8
Method Wrapping Process

• Simulink wrappers for ITK, VTK
• Each Simulink block is an S-Function with
  callbacks
• S-Function callbacks are C code, calls ITK or
  VTK
• CMake build process
• Start with an XML description of each filter
  (ITK)
• or Start with class header files (VTK)
• CMake calls perl script to generate C code
• CMake then calls perl to generate .mdl library
  file
• Finally, build libraries with VC or gcc

9
Method Data Flow for SimITK

• SimITK, SimVTK use different architecture
• ITK is primarily image based
• VTK handles many data set types
• SimITK Data Flow
• ITK images -gt MATLAB arrays (info and data)
• Use shared memory for efficiency
• Conversion of ITK parameter data types
• ITK uses several specialized types e.g. versor

10
SimITK Helper Class
Input
InputPort
Conversion
itkImage
Spacing Origin
Data
Conversion
Parameters
Run ITK Image Filter
Parameter Variables
Output
Conversion
OutputPort
Spacing Origin
itkImage
Data
11
Method Data Flow for SimVTK

• SimVTK Data Flow
• Use MATLAB arrays to pass vtkObject pointers
• Define VTKOBJECT MATLAB type for safety
• When run starts, Simulink builds VTK pipeline
  on-the-fly
• Then, Simulink calls Update or Render at each
  time step
• SimVTK filter parameters
• VTK parameter types need no conversion
• VTK and MATLAB both use basic C data types

12
Registration Examples

• Registration in ITK
• Immediate, interactive feedback
• Great way to tune parameters
• Easy to swap out components

13
(No Transcript)
14
Intervention Examples

• Computer Assisted Intervention in VTK
• Incorporate real-time data input
• Use Subsystems to manage complexity

15
(No Transcript)
16
(No Transcript)
17
Discussion

• Diagrams that actually run as code
• Easy to verify that code matches design
• Great teaching tool
• Understandability improves maintainability
• Utility
• Simulink models can run as stand-alone exec
• Nearly all VTK, ITK classes can be wrapped

18
Future Work

• WrapITK
• To automate ITK build (VTK is already done!)
• GUI integration
• Slicer module interface in Simulink
• KWWidgets (later Qt, maybe FLTK)
• Community Building
• Mailing list, website
• We want to hear from you!