Validation of Defined Regions of Interest Using Surface Scanning

Objective:

• To evaluate manual and automated segmentation routines using surface scans of disected bones from cadaveric specimens

Progress:

• CT images of the hand and wrist have been obtained
• Four hands have been disected and bones extracted
• Surface scanning for these disected bones is completed using a Roland LPX-250 Laser Scanner
• Manual tracing of the CT images is completed
• Evaluate reliability of the manual tracers is complete using relative overlap
• Tools have been developed to reorient surface axes and to register the manually defined model and the surface resulting from laser scanning
• Distance maps for the manual segmentation have been generated.
• The effect of various post processing routines have been studied relative to changes in the surface geometry relative to the laser scan.

To Do:

• Evaluate automated segmentation algorithms to the Physical laser scans
  1. Neural network segmentation
  2. EM Segmentation Module
• Integrate surface orientation and registration tools into Slicer3
• Integrate surface distance map tools into Slicer3 - These could be useful for other applications as well.

Key Investigators:

• Iowa: Nicole Grosland, Vincent Magnotta, Nicole DeVries

Links:

• Musculoskeletal Imaging, Modelling and Experimentation (MIMX)

References:

• Devries NA, Gassman EE, Kallemeyn NA, Shivanna KH, Magnotta VA, Grosland NM. Validation of Phalanx Bone 3D Surface Segmentation from CT Images Using Laser Scanning, Skeletal Radiology, 2007

Figures:

Surface distance mapping tool developed in FLTK

Surface Distance Map for a phalanx bone between a surface manually defined via CT and a physical surface laser scan.

Overlay of the physical surface (White) and CT based surface (Purple)