Differences between revisions 5 and 6
Revision 5 as of 2015-11-09 18:56:22
Size: 3243
Editor: aerdemir
Comment:
Revision 6 as of 2016-01-25 19:56:25
Size: 6253
Comment:
Deletions are marked like this. Additions are marked like this.
Line 48: Line 48:
== Results ==

Following are examples (oks001) of distance maps (using !MeshLab: Hausdorf distance) between various pairs of surfaces to assess:

 1. distance between different model components (i.e. femur vs. femur cartilage, BACK SURFACE)
 1. spatial error between corresponding surfaces with varying levels of smoothing/refinement
 1. thickness of model components (i.e. cartilage):

To map tissue thicknesses (e.g. cartilage), opposing surfaces (i.e. BACK SURFACE, FRONT SURFACE) from a manifold (i.e. watertight) STL mesh were split into two complementary, shell STL mesh sets using Blender.

=== Distance (mm) Between Model Components ===

 FEMUR vs. FEMUR CARTILAGE:
 * femur (REF; VCG smoothed, 0.7; ISO, 10) vs. femur cartilage, BACK SURFACE (VCG smoothed, 0.7; ISO, 10):

  [[ImageLink(oks001_FMB_VCG_0.7_ISO_10_vs_FMC_SAME_AI.png, width=300, alt=oks001_FMB_VCG_0.7_ISO_10_vs_FMC_SAME_AI)]]

 FEMUR CARTILAGE vs. FEMUR:
 * femur cartilage (REF; VCG smoothed, 0.7; ISO, 10) vs. femur (VCG smoothed, 0.7; ISO, 10):

  [[ImageLink(oks001_FMC_VCG_0.7_ISO_10_vs_FMB_SAME.png, width=300, alt=oks001_FMC_VCG_0.7_ISO_10_vs_FMB_SAME)]]

=== Smoothing Errors (mm) ===

 UNSMOOTHED vs. SMOOTHED:
 * femur cartilage, BACK SURFACE (REF; unsmoothed) vs. femur cartilage, BACK SURFACE (VCG smoothed, 0.7; ISO 10):
 * femur cartilage, FRONT SURFACE (REF; unsmoothed) vs. femur cartilage, FRONT SURFACE (VCG smoothed, 0.7; ISO 10):

  [[ImageLink(oks001_FMC_vs_FMC_VCG_0.7_ISO_10.png, width=300, alt=oks001_FMC_vs_FMC_VCG_0.7_ISO_10)]]

  [[ImageLink(oks001_FMC_vs_FMC_VCG_0.7_ISO_10_INF.png, width=300, alt=oks001_FMC_vs_FMC_VCG_0.7_ISO_10_INF)]]

=== Tissue Thickness (mm) ===

 FEMUR CARTILAGE THICKNESS (VCG smoothed, 0.7):
 * femur cartilage, BACK SURFACE (REF; VCG smoothed, 0.7) vs. femur cartilage, FRONT SURFACE (REF; VCG smoothed, 0.7):
 * femur cartilage, FRONT SURFACE (REF; VCG smoothed, 0.7) vs. femur cartilage, BACK SURFACE (REF; VCG smoothed, 0.7):

  [[ImageLink(oks001_FMC_VCG_0.7_BACK_vs_FMC_SAME_FRONT.png, width=300, alt=oks001_FMC_VCG_0.7_BACK_vs_FMC_SAME_FRONT)]]

  [[ImageLink(oks001_FMC_VCG_0.7_FRONT_vs_FMC_SAME_BACK_INF.png, width=300, alt=oks001_FMC_VCG_0.7_FRONT_vs_FMC_SAME_BACK_INF)]]

 FEMUR CARTILAGE THICKNESS (VCG smoothed, 0.7; ISO, 10):
 * femur cartilage, BACK SURFACE (REF; VCG smoothed, 0.7; ISO, 10) vs. femur cartilage, FRONT SURFACE (REF; VCG smoothed, 0.7; ISO, 10):
 * femur cartilage, FRONT SURFACE (REF; VCG smoothed, 0.7; ISO, 10) vs. femur cartilage, BACK SURFACE (REF; VCG smoothed, 0.7; ISO, 10):

  [[ImageLink(oks001_FMC_VCG_0.7_ISO_10_BACK_vs_FMC_SAME_FRONT.png, width=300, alt=oks001_FMC_VCG_0.7_ISO_10_BACK_vs_FMC_SAME_FRONT)]]

  [[ImageLink(oks001_FMC_VCG_0.7_ISO_10_FRONT_vs_FMC_SAME_BACK_INF.png, width=300, alt=oks001_FMC_VCG_0.7_ISO_10_FRONT_vs_FMC_SAME_BACK_INF)]]

 NOTE: spurious distances (0.0 mm, which are known not to be correct) are present only in the FRONT SURFACE (REF) vs BACK SURFACE!!!

TableOfContents()

Target Outcome

Geometric reconstruction of a tissue of interest ready for meshing:

  • as a (parametric or explicit) watertight surface representation, or
  • as a solid geometry

Prerequisites

Include(Infrastructure/AuxiliarySoftware, "Infrastructure", 2, from="= Geometry & Mesh Generation & Manipulation =", to="= Scripting & Numerical Analysis =")

Previous Protocols

For more details, see ["Specifications/ImageSegmentation"].

Protocols

Input

  • Volume of tissue of interest as a binary image aligned with original MRI coordinate system (raw, without filtering and smoothing)
  • Surface representation of tissue of interest in STL format in MRI coordinate system (raw, without filtering and smoothing)

Procedures

-- ["aerdemir"] DateTime(2013-12-30T15:14:36Z) This section may list general purpose procedures and alternatives to utilize the same input to reach the same output. Procedures optimized for specific tissues should also be provided in here.

When starting with volume representation

  • Confirm the tissue volume is a raw representation (without filtering and smoothing)
  • Generate raw surface representation of the tissue volume
    • Upload tissue volume in image segmentation software, e.g. Slicer.
    • Generate water-tight triangulated surface representation (no filtering, smoothing).
    • Save raw surface representation with the same file name as the volume representation except with the .stl extension.
  • Continue with the next section.

When starting with surface representation

  • Load raw surface representation of the tissue volume in MeshLab.

  • Smooth raw surface representation, e.g.:
    • Under Filters->Smoothing->Laplacian Smooth, Select smoothing steps of 9 and uncheck all other boxes and apply. This will smooth down most jagged edges from the original file

    • Under Filters->Smoothing->Two Step Smooth, Select smoothing steps of 3, Feature Angle Threashold at 60, Normal smoothing steps at 20, and Vertex Fitting steps at 20 and apply. This step further levels uneven edges

    • Under Filters->Smoothing->Taubin Smooth, Select Lambda to 1, mu to 0.0, and Smoothing steps to 12, all boxes are unchecked and press apply. This step creates an overall uniform surface with few rough edges, while still keeping the shape of the initial object. Some values may vary depending on the model input, this selection works for large bones like the femur.

  • Parametrize and resample smoothed surface representation.
  • Save smooth surface representation with the same file name as the volume representation except with trailing text of 'smooth'.

When parametric geometry is needed

Results

Following are examples (oks001) of distance maps (using MeshLab: Hausdorf distance) between various pairs of surfaces to assess:

  1. distance between different model components (i.e. femur vs. femur cartilage, BACK SURFACE)
  2. spatial error between corresponding surfaces with varying levels of smoothing/refinement
  3. thickness of model components (i.e. cartilage):

To map tissue thicknesses (e.g. cartilage), opposing surfaces (i.e. BACK SURFACE, FRONT SURFACE) from a manifold (i.e. watertight) STL mesh were split into two complementary, shell STL mesh sets using Blender.

Distance (mm) Between Model Components

Smoothing Errors (mm)

Tissue Thickness (mm)

Output

  • Explicit surface representation of tissue of interest; in STL format in MRI coordinate system (processed with volume preserving smoothing)
  • Parametric surface representation of tissue of interest in IGES & STP format in MRI coordinate system (processed with volume preserving smoothing)

  • Parametric solid model of tissue of interest in IGES & STP format in MRI coordinate system (processed with volume preserving smoothing)

Specifications/GeometryGeneration (last edited 2020-10-13 13:59:46 by klonowe)