Docstoc

NA-MIC_Surface-Scan

Document Sample
NA-MIC_Surface-Scan Powered By Docstoc
					NA-MIC
National Alliance for Medical Image Computing
http://na-mic.org



Validation of Bone Models
Using 3D Surface Scanning

Nicole M. Grosland
Vincent A. Magnotta
Validation
• True “gold-standard” often very
  difficult to achieve
      – Brain imaging often have to live with
        manual raters
      – Established guidelines based on
        anatomical experts
• Are there better “gold-standards” for
  other regions of the body?

National Alliance for Medical Image Computing
http://na-mic.org
Orthopaedic Imaging
• Ideas developed out of goal to automate
  the definition of regions of interest of the
  upper extremity
      – How can we validate these automated tools?
• Orthopaedic applications it is possible to
  dissect the region out of cadeveric
  specimens
      – Use specimen itself as the “gold-standard”



National Alliance for Medical Image Computing
http://na-mic.org
3D Laser Scanner
• 3D Laser scanners have been used
  for rapid prototyping and to non-
  destructively image ancient artifacts
• Roland LPX-250 Scanner Obtained
      – Planar and rotary scanning modes
      – 0.008 inch resolution in planar mode
      – Objects up to 10 inches wide and 16
        inches tall can be scanned
      – Reverse modeling software tools
National Alliance for Medical Image Computing
http://na-mic.org
LPX-250 Laser Scanner




National Alliance for Medical Image Computing
http://na-mic.org
Specimens
• 15 cadaveric specimens were
  obtained spanning the distal radius
  to the finger tips
      – Specimens mounted on a Plexiglas
        sheet in the neutral position
• CT images collected on a Siemens
  Sensation 64 scanner
      – Images obtained with a 0.2x0.2x0.4mm
        resolution

National Alliance for Medical Image Computing
http://na-mic.org
Image Post Processing
• Resampled images to 0.2mm3 resolution
      – Images cropped at the wrist
• Manually defined the proximal, medial,
  and distal phalanx bones
      – Two raters defined these regions on 11 fingers
      – Inter-rater reliability evaluated using relative
        overlap (0.91, 0.90, and 0.87 respectively)
      – Surfaces from the binary masks were
        generated


National Alliance for Medical Image Computing
http://na-mic.org
CT Scan




National Alliance for Medical Image Computing
http://na-mic.org
Finger Dissection
• Phalanx and metacarpal bones removed
      – Care taken to avoid tool marks on the bones
• De-fleshing process outlined by Donahue
  et al (2002) was utilized
      – Bones allowed to soak in a 5.25% sodium
        hypochlorite (bleach) solution for 6 hours
• Degreased via a soapy water solution
• Thin layer of white primer was used to
  coat the bony surfaces

National Alliance for Medical Image Computing
http://na-mic.org
                                              Prepare Specimen for Scanning
                                              Deflesh, Degrease, Paint, Embed in Clay



Scan Distal End Using Dr.                                                                                 Scan Proximal End Using Dr.
PICZA 1-Plane Scanning                                                                                      PICZA 1-Plane Scanning

                        Scan Long Axis (Distal End Up) Using              Scan Long Axis (Proximal End Up)
                            Dr. PICZA 4-Plane Scanning                    Using Dr. PICZA 4-Plane Scanning


                                                 Edit the Scanned Surface Using Dr.
                                                        PICZA Editing Tools
                                                Remove Noise, Delete Abnormal Faces, Create
                                                              Polygon Mesh




                                           Use Pixform Software to Further Edit Surface
                                          Delete Extraneous Vertices, Fill Holes in Surface, Clean Non-
                                                          manifold and Crossing Faces




 Align, Register, and Merge the Long Axis (Distal                                        Align, Register, and Merge the Long Axis
           End Up) and the Distal End                                                   (Proximal End Up) and the Proximal End



                                             Align, Register, and Merge the Distal and
                                                    Proximal Ends of the Bone


                                         Smooth Final Surface with a Tolerance of 0.10 mm
            National Alliance for Medical Image Computing
            http://na-mic.org
                        Proximal Bone Surface Scanning Steps
                                       Specimen CA05042125L




                   Distal End                                  Proximal End



Distal Up                                                                           Proximal Up




            Distal Merge                                                Proximal Merge




                                                                      Full Finger Scan


               National Alliance for Medical Image Computing
               http://na-mic.org
                  Proximal Bone – CA05042125L




National Alliance for Medical Image Computing
http://na-mic.org
Middle (Green) and Distal (Pink) Bones – CA05042125L




National Alliance for Medical Image Computing
http://na-mic.org
Full Finger Surface Scans




          Full Finger –                         Full Finger –
          CA05042125L                           MD05010306R




National Alliance for Medical Image Computing
http://na-mic.org
       Full Finger –
       MD05042226L


                                                Full Finger –
                                                SC05030303R




National Alliance for Medical Image Computing
http://na-mic.org
 Registration of Surfaces
• Surface scans origin shifted to center
  of mass and reoriented to have the
  same orientation as the CT data
• Surfaces registered using a rigid
  iterative closest point algorithm
• Compute Euclidean distance
  between the surfaces


 National Alliance for Medical Image Computing
 http://na-mic.org
Surface Distance
Measurement Tool




National Alliance for Medical Image Computing
http://na-mic.org
      Surface Distances: 1P-SC05030303R




Proximal Distance Map
                                                     Laser scanned surface
                                                     Traced surface


     National Alliance for Medical Image Computing
     http://na-mic.org
   Surface Distances: 1M-SC05030303R




Middle Distance Map
                                                    Laser scanned surface
                                                    Traced surface


    National Alliance for Medical Image Computing
    http://na-mic.org
Surface Distances: 1D-SC05030303R




   Distal Distance Map                          Laser scanned surface
                                                Traced surface

National Alliance for Medical Image Computing
http://na-mic.org
Results
                                                Average
                                                Distance
                      Finger ID
                                                between
                                                Surfaces
               1P-SC05030303R                    0.145

               1M-SC05030303R                    0.134

               1D-SC05030303R                    0.142

               1P-MD05010306R                    0.142

                Average values                   0.142

National Alliance for Medical Image Computing
http://na-mic.org
Discussion
• Surface scans used to validate regions of
  interest generated via CT scans
      – Average distance less than 1 voxel (0.2mm)
• Surface scans can be used to evaluate
  image processing procedures
      –   Validation of tracing guidelines
      –   Amount of smoothing
      –   Iso-surface threshold
      –   Evaluation of automated segmentation
          routines

National Alliance for Medical Image Computing
http://na-mic.org
Future Work
• Make regional specific measurement
      – E.g. articulating surface
• Evaluate ANN segmentation using
  this technique
• Can this be used to evaluate soft
  tissue geometry?



National Alliance for Medical Image Computing
http://na-mic.org
Thanks
• Nicole Grosland – Project PI
• Esther Gassman
      – Manual Traces/Surface scanning
• Nicole Kallemeyn
      – Manual tracing and surface comparison
• Nicole Devries
      – Finger dissection and specimen prep
• Kiran Shivanna
      – Software development
• Stephanie Powell
      – Automated segmentation

Work funded in part by NIH/NIBIB Grant 1R21EB001501-01A2



National Alliance for Medical Image Computing
http://na-mic.org

				
DOCUMENT INFO