1. Visual Optimality And Stability Analysis Of 3DCT Scan Positions Artem Amirkhanov 1,2 Michael Reiter 2 Christoph Heinzl 2 M. Eduard Gröller 1 1 Institute of Computer Graphics and Algorithms Vienna University of Technology 2 Upper Austrian University of Applied Sciences Wels Campus, Austria

2. Artem Amirkhanov 1 Scanning Geometry X-ray source Specimen Detector Rotary plate

3. Placement is crucial for Measurements accuracy Iso-surface quality Requires knowledge, intuitive Good placement is hard to find For a complex specimen For measurement features Doing several scans is expensive Artem Amirkhanov 2 Specimen Placement

4. CAD model is available Quality control – from manufacturer Reverse engineering – preliminary scan Use a CAD model to estimate a good placement Visual analysis tool is required for Simulation-based preview Optimal placement estimation Tool Artem Amirkhanov 3 Motivation Optimal Placement

5. Artem Amirkhanov 4 Why Placement is Important? Beam-hardening artifacts Cupping artifacts Streaks Bad planar faces artifacts Blurring

6. Artem Amirkhanov 5 Good/Bad Placement Example Bad: Good:

7. Shortest penetration lengths Method: ray casting Smallest bad planar faces area Method: Radon-space analysis Stable within a certain range of reliability Method: stability widget Artem Amirkhanov 6 Criteria and Methods

8. Artem Amirkhanov 7 Workflow Ray casting Radon-space analysis Simulation Stability analysis Data exploration and analysis Visual analysis Optimal placement CAD model Parameter spaces

9. Artem Amirkhanov 8 Simulation Set of candidate placements Placement is defined by the orientation Orientation is defined by two Euler angles α and β α β

10. Artem Amirkhanov 9 Ray Casting Simulation Reflects the scanning setup Set of projections for every placement Parameters Maximum penetration length Average penetration length

11. Artem Amirkhanov 10 Radon-Space Analysis Supporting planes → points in Radon space Finding bad points in Radon space Parameter Bad faces area percentage

12. Artem Amirkhanov 11 Parameters Representation Best Worst Parameter value:

13. Artem Amirkhanov 12 Stability Widget Better SameWorse Parameter value:

14. Penetration-length histograms Visualizing ray subsets. Color coding bad faces. Artem Amirkhanov 13 Data Visualization and Exploration

15. Artem Amirkhanov 14 Evaluation Using Dimensional Measurements Two test-part specimens Measurement features Radius Distance CALYPSO used to estimate measurement quality

16. Complex real world specimen Best and worst placements Variance comparison of the iso-surfaces with the CAD model Artem Amirkhanov 15 Evaluation Using Variance Comparison

17. Artem Amirkhanov 16 Evaluation Bad PlacementGood Placement Position deviation spectrum in [mm] 0 -221 0 -221

18. Artem Amirkhanov 17 Limitations and Future Work Orientation only Detect orientation and position in sequence Still requires user interaction Combining parameters

19. Artem Amirkhanov 18 Conclusions Thank you! VS Contact: artem.ogre@gmail.com