New way for minimum cross-section area. Instead of setting limits and drawing cross-sections, then draw points down the airway, then automatically picks out areas and color codes them. Showed how to show the sinuses.
Can superimpose volumes, e.g. pre and post surgical advancement of jaws. Voxel based superimposition and regional voxel based superimpoisition. Do a rough superimposition by identifying points and then superimposing. Better yet do regional superimposition, get better superimposition by doing voxel based. If using different machines, voxel sizes don’t have to be the same size to do this.
Can save individual airway files as .stl files and then do superimposition. Will check with Swan about how to do more than two. To export .stl, go from photo display
2-sided sensitivity adjustment to identify areas in range between two houndsfield units.
Can measure both volumes also as long as boundaries set same way.
Ability to export T1 into same coordinates in T2?
Color map of distance between stl and dicom
Automatic algorithm to superimpose .stl to dicom
Can export several (e.g. both dental arches and airway) in cone beam coordinate system into Meshlab, 3d party software.
To do photo wrap with colors onto cone-beam – digitize four mid-sag points on face on both images.
Can do overlay superimposition of 3d photo over conebeam. Do approximate, then do autosuperimpose.
Showed how to do osteotomies.
Since Dec 2014, splints can be exported in .stl format.
Custom structures should be assigned to cranial base, maxilla, mandible areas so that they would transfer to superimpositions.
3D surgery software is available to school. Can compare predicted surface over actual post-tx surface.
Will ask Swan about regional surface to surface or surface to volume superimposition
___________
Jhimli Mitra, Case Center for Computerized Imaging and Personalized Diagnostics, School of Engineering
Brain atlas
Need method for automatic segmentation of regions or volumes without manual intervention.
Need alignment (registration) of volumes or images of subjects into a common frame.
Size doesn’t matter, do scaling with rigid transformation so voxels map. For same patient longitudinally, do affine. Between people (average brain atlas to person) need to transform voxel size.
Registration – affine -- adjusts for rotation, translation and scale – bony structures
Deformable – adjusts for local deformation at pixel or voxel level o—soft tissue structures
Use voxel wide method for registration and voxel-wide statistical analysis
Brain tissue segmentation
Estimate volume of white matter and gray matter and cerebrospinal fluid in a probabilistic tissue
Skull-stripping on brain MRI, do deformable registration of atlas on the patient’s face, apply transform and it’s automatic.
Cortical and subcortical parcellations, e.g. change in hippocampus volume.
They’re doing white-matter tract connectivity. They do deformation analysis for gliolastoma multiforme linking deformation to survival.
Brain atlases are freely available witwh different kinds of parcellations. MNI152, AAL 116
Do growing brains in separate atlases, not mixed with adults. Age specific atlases are available.
Could do something similar for face atlases. Dolphin has tools for 2D averaging.
_______________
Dave Schecter (formlabs.com) - 3D Printer demo as google meeting
Form 2 printer kit $3500, 1 liter standard resin $150
Desktop SLA printer, stereolithography machine - Liquid resin in tray, hardened by laser, differs from fused deposition printers
Build platform 5.7 inches X 5.7 inches X 7 inches tall
Finishing kit – was printout with isopropyl alcohol.
All resins paintable
Liquid resin stable 6 months, comes in three consistencies from tough to soft
_______________________
Motionview – Dan Knoch
3D printer
Indirect bonding software
Intraoral scanners all make bad data with holes. Takes too long.
_____________________
February 18th
Comparison of Dolphin and Slicer superimpositions
Case thesis – Mohammed Bazina
Slicer allows registration on irregular shape. Dolphin must use box.
Results are very similar.
Slicer method: Manual approximation first, segment cranial base, use Slicer to register on the cranial bases (takes about 15 files to do this). Runs on Macs.
Dolphin – use landmarks for approximation with one in mandible to minimize rotation. Set boundary boxes. Then do voxel base superimposition. Dolphin now allows T2 to be exported with coordinates of T1. Then DICOM is segmented and compared in Slicer to allow the investigato to find differences in the Slicer and Dolphin skulls. Differences between registered T2s is about .25mm. To get numbers for statistics, picked 7 areas for measurements. Gonial angles have the biggest differences. Minimum area is nasion (part of cranial base registration).
Conclusion: Dolphin is faster, easier, but needs color map and has purchase cost.
Discussion: Non growing subjects. Choose cranial nerves uptimately, or maxilla or mandible.
Slicer allows freeform registration area, Dolphin box may be inadequate for maxilla or mandible.
Right now can save time doing registsration in Dolphin and then export to Slicer for color map.
Discussion: Is cranial base stable in transverse dimension? Off-midline fiducials?
KAU - Ultrasound based jaw tracking
Fiducials in maxillary tray, open in hinge, protrusive, lateral, chewing
Segment the conebeam. Information is saved in ultrasound system and integrated.
Omniscan taken of dentition, make .stl casts
Showed examples of surgical patients.
_____________
topsCephMate demo: Toni Magni
Implement mean change expansion by Lysle Johnston? Rendering unto Caesar also.
Incorporate as mean and minimum/maximum (what could happen and what is likely).
Dolphin: Ken Gladstone
For conebeam derived lateral cephs, can use F8 to switch between renderings to help locate points.
The voice is named “Margaret.”
Ortho2: Mark Hans
has Bolton standard overlay built in and it can be saved.
________________________
Discussion about advantages of 3D facial scanning
No radiation, multiple time points, superimposition, better details of treatment outcomes
_______
Auto-superimposition of two images of same skull which has metal markers
Dolphin is 32 bit, limited to 2G memory.
3D mandibular superimposition -- Do symphysis/pogonion voxel superimposition, click done, back to single image and put center of rotation at point on chin, rotate to third molars. Show both nerve canals in image. Do they save in vol/vol superimposition? Not yet.
Tried superimposition surface to volume of brain image to 17 year old patient.
Group will compile JCEG registry of existing CBCT images.
________
Digital study casts for customized appliances, Dominico Dalessandri, University of Bresia
Virtual set-ups using CBCT data
Lit review JCO 2011 45:209-216
Mimics & Dolphin versus 3Shape
Keep roots in bone
Intermediate steps difficult
3 Shape doesn’t allow slices anymore, but can do all intermediate steps. Software instability, indirect DICOM control
Manual segmentation required
Dan Knoch does segmenting of scanned crowns and places in CBCT with template roots. Template roots are deformable to shape of natural root. (tracing in 3D) Intraoral scans accurate to 30-40 microns. CBCT only accurate to 300-400 microns.
Ken Glandstone says it is possible right now to trace incisors and molars as lines.
Duterloo book on 2D cephalometrics/superimpositions
Mark Hans will give Dolphin male and female Bolton templates to Dolphin and Tops. Case does not have the vector graphics – that was done by Ortho 2.
JADA, Demster and Duddles, 1964 – average roots
About the atlases – maybe import scans of same subjects dental cast rather than averaging teeth on CBCTs.
Invisalign accepts Itero, 3m trudef (incognito) and Cerec (Sirona). Main Invisalign patent expires this year. Cerec Omnicam (Patterson) and Dolphin (Pattersono) are tightly integrated. It’s color and powderless. Dolphin takes any .stl files.
Face Scanning – captures different information ompared to cone beams. Cannot replae. However, are useful because longitudinal studies can be done, i.e. Face Base. Would be especially useful for applications that need to assess soft tissue changes. Per Johan Wisth, University of Bergen
Database of 3D CT images, create a registry of CBCT users
Johnston forecast grid is useful and probably should be implemented in software.