Virtual CT acquisition and reconstruction of complex and ... · Complex and noisy scanning trajectories. 13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy

9th Conference on Industrial Computed Tomography, Padova, Italy (iCT 2019)

Virtual CT acquisition and reconstructionof complex and noisy scanning trajectories in aRTist

Carsten Bellon, Karin Burger, Christian GollwitzerBundesanstalt für Materialforschung und -prüfung (BAM), Unter den Eichen 87,

12205 Berlin, Germany, e-mail: carsten.bellon@bam.de

Abstract

In modern CT imaging, simulation has become an important tool to minimize cost- and time-intensive measurements. It is increasingly used to optimize techniques for complex applications, to support the preparation of written procedures, and for educational purposes. We extended the CT simulation software ‘aRTist’ with a module to set-up arbitrary trajectories including disturbing influences during the scan. Moreover, such geometric deviations can be compensated by the internal reconstruction tool.

Keywords: simulation, arbitrary trajectories, projection matrix, filtered back projection

VIRTUAL CT ACQUISITION AND RECONSTRUCTION OF COMPLEX AND NOISY SCANNING TRAJECTORIES IN ARTIST

Carsten Bellon, Karin Burger, Christian Gollwitzer

13.02.2019

Outline

Virtual CT with aRTist

Complex scanning trajectories

Acquisition and reconstruction with individual projection matrices

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 2CC DavidMCEddy at wikipedia

George Box1919 - 2013

“Essentially, all models are wrong, but some are useful.“Box, George E. P.; Norman R. Draper (1987). Empirical Model-Building and Response Surfaces, p. 424, Wiley. ISBN 0-471-81033-9

17IND08 AdvanCTwith one of the objectives:Uncertainty estimation using virtual CT

Virtual CT with aRTist

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 3

Simulator aRTist

Design objectives for aRTist

‒ Providing RT modeling software for practitioners

‒ Easy to use user interface

‒ Containing models for all relevant RT components

‒ Handling complex part geometries

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 4

Simulator aRTistSource parameters, setup with preview image

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 5

Simulator aRTistPump cover (Al) overlaped by an box (air)

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 6

Simulator aRTistAdditional ball (Al) overlapping part and box

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 7

Simulator aRTistScatter image and selected photon traces (Monte Carlo)

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 8

Simulator aRTistComputerized Tomography

Virtual CT setup

Number of Projections

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 9

Simulator aRTistComputerized Tomography

Scatter imagePrimary image

Profile plot

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 10

Without scatter contribution

With scatter contribution

Complex scanning trajectories

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 11

Complex and noisy scanning trajectories

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 12

‒ Geometric modification functions for each part in the radiographic scene

‒ Variation types: linear, logarithmic, circular,… , random, user defined

Module TomoSynth supports easy setup of series of repetitive projections with deterministic and random variations

‒ Laminography, helical CT

‒ Deviations from ideal CT scan trajectories

TomoSynth moduleSimulation of helical CT

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 13

Modification functions

‒ Source and detector:angle linear: from 0 to 16*360theta constant: 90

‒ Part and support:X position linear: from 0 to 200

TomoSynth moduleCone-beam CT (ideal)

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 14

Modification functions

‒ Part:angle linear: from 0 to 360theta, phi constant: 90

TomoSynth moduleCone-beam CT w. instable rotation axis

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 17

Modification functions

‒ Part:angle linear: from 0 to 360theta constant: 90phi gaussian: mean 90 stddev 2

Reconstruction with individual projection matrices

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 18

Reconstruction with individual projection matrices

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 19

Standard reconstruction tool: a single projection matrix is considered containing the geometric information of the setup

Extended reconstruction tool: individual projection matrices are used for every angular step to allow for a more accurate reconstruction in case of deterministic motion

Modification functions

‒ Source and detector:angle linear: from 0 to 360 (circular scanning)

‒ Source:Z shift formula: 3.0 * sin(30.0 * $step / $Nsteps)(sinusoidally moving in Z)

Projection matrices are stored for every angular step

Virtual CT of arbitrary trajectoriesExample: sinusoidal source movement

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 20

Virtual CT of arbitrary trajectoriesExample: sinusoidal source movement

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 21

source shift parallel to the rotation axis formula: 3.0 * sin(30.0 * $step / $Nsteps)

Example: sinusoidal source movement Reconstruction result

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 22

The image resulting from a standard reconstruction shows strong artifacts from the sinusoidal source movement (left)

Single projection matrix Ideal scan, without extra source movment

Example: sinusoidal source movement Reconstruction result

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 23

Individual projection matrices allowed to take the known movement into account, which results in a significant reduction of artifacts (right)

Single projection matrix Individual projection matrices

… without any additional computational effords!

Conclusions

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 24

Conclusions

The radiographic modeling software aRTist can effectively used for virtual CT from standard cone-beam to complex scanning trajectories.

Deviations from ideal CT scan trajectories can be easily simulated which is a necessary step towards uncertaintydetermination from simulation.

With the possibility of storing one projection matrix per angle, the impact of including deterministic motion in the reconstruction process can now be estimated.

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 25

Further Information

www.aRTist.bam.de

13.02.2019 Virtual CT acquisition and reconstruction of complex and noisy scanning trajectories in aRTist 26