1. Software for Neutron Imaging Analysis
A.P. Kaestner
Software for Neutron Imaging Analysis
Imaging and Engineering STAP (Fall 2018)

2. What happens after an imaging experiment?

3. Data processing can be a very labor- and computationally intense task
Basic processing
Minutes, hours
CT reconstruction
Image processing
Hours
Visualization
Analysis
Hours, days
Kaestner et al, 2012
Woracek et al, 2014
Gruenzweig et al, 2012

4. Analysis workflow
Needs creative interaction
Can be generalized

5. Data evaluation fitness of users
Users neutron imaging have very different background and experience

6. Data evaluation fitness of users - Characterization
Users neutron imaging have very different background and experience
Beamline scientist
Early stage PhD student
Cultural heritage users
Serving people with different background and needs is a great challenge

7. Computing infrastructure needed
Laptop
Workstation
Cluster
Pre-evaluation
Training sessions
Small data
Most evaluation
Explore data
Mid sized data
Mass evaluation
Full sized data
Software needs to be scalable to support different infrastructure

8. Different parts of the development
Core
Physical models
Fitting
Geometry
Support algorithms
Segmentation
Data base access
Interface external tools
Image analysis
File handling
GUI
Interaction
Provide parameters
Visualization UX
CLI
Scripting support
Mass processing

9. Applications developed at PSI – White beam
MuhRec
CT reconstruction tool
Configurable processing chain
KipTool
General processing tool for 3D images
Converter tools
NIQA tool (not IKC)
Analysis of IAEA QA samples
nGI tool
Reduction of phase stepping scans

10. Work in progress - Energy resolved imaging
New neutron sources with pulsed beam …
… Reveal new information about the sample  New science
Bragg edge analysis
Diffraction imaging
Modulation techniques
Challenges on the development
Produce large data
Higher dimensional data
Workflow flexibility
Visualization options

11. Manpower Manpower according to TA Steering committee: Time gap until
W. Kockelmann, R. Woracek, M. Strobl, M. Lerche
Time gap until
hot commissioning

12. Modern software development
Agile development
Iterative development
Collaboration
Using repositories
New issues in repository branches
Maintain stable master through merge reviews
Issue tracking (New features, Improvements, Fixing bugs)
Frequent releases (2 x year)
Automated testing and builds
Online documentation

13. Some technical details about the development
Main coding language
GUI Library
Planned binding
In particular: C++11
XCode (Mac)
MSVC15 (Windows)
g++ (Linux)
Version: 5.9 LTS
Cross platform
Qmake
Qtest
Installer
Python 3
Embedded scripting
Call libs from python
Tutorials
Typical work flows
Repository: GitHub
Development history
Issue tracking
Documentation (Wiki)
Build server: Jenkins
Automated builds
Planned:
Nightly build installers
Automatic unit testing

14. Collaboration Network
PSI
ESS ERIC
TUM
SNS
ISIS
J-PARC
Contributors
Observers

15. Training and documentation
Direct user instruction during experiments
Hands-on sessions
Varenna Imaging Workshop 2017
WCNR 11 – Pre-conference school
SwedNESS imaging PhD School
User documentation online
Tutorial videos on YouTube

16. Wir schaffen Wissen – heute für morgen
The analysis of imaging data should be …
… done in a reproducible way.
… open source to promote collaboration and reproducibility.
… developed with modern techniques.