Partners: TU Delft, VU, CWI

1
SP2.3 UI and VR Based Visualization

• Partners TU Delft, VU, CWI

• Ongoing Activities and progress
• Collaboration Highlight with SP 1.6 DUTELLA

R. van Liere April 7th, 2006
2
SP 2.3 people

• 4 PhD students
• Broersen, Burakiew, Kruszynski (CWI)
• van der Schaaf (VU)
• 3 PD
• Botha, Koutek (TUD)
• de Leeuw (CWI)
• 4 supervision
• van Liere
  (CWI)
• Post, Jansen (TUD)
• Bal
  (VU)

3
SP2.3 ongoing activities

• Multi-spectral visualization SP 1.6
• Particle visualization SP 1.6
• Confocal Cell Imaging
• Volume measuring SP 2.1
• Medical Imaging SP 1.4
• Virtual Reality on the GRID SP 3.1
• Distributed Scene Graphs SP 3.1

4
SP 2.3 status

• 25 international publications
• 2 spin-offs
• Foldyne (TU Delft)
• Personal Space Technologies (CWI)
• Projected output
• 4 PhD thesis
• At least 2 packages in PoC

5
Collaboration SP 1.6 DUTELLA

• Prof Ron Heeren (ALMOF)
• Topic Mass Spectrometry for molecular imaging
• Motivation need for better MS analysis tools
• Visualization Topics
• Multi-spectral data visualization
• In-silico mass spectrometry
• Envisioned output
• GRID enabled toolbox for MS analysis
• Applications according to VL-e methodology

6
Problem aligning multi-spectral data cubes

• Multi-spectral data cube 256x256x65k
• Multiple data cubes
• 100 cubes in mosaic
• Current procedure manual alignment on pixel
  values

7
Our novel approach

• Idea Align spectral features in adjacent samples
• Approach
• Compute spectral features using PCA
• For each feature, find a most optimal spatial
  alignment of the feature
• The overall spatial alignment is optimal for all
  features

8
MS beelden dijbeen muis
9
First Spectral Feature
Principal Component1
10
Second Spectral Feature
Principal Component2
11
Minima landscape
Minimization map of 1st feature
12
Impact ? Generic ? GRID?

• Faster, unsupervised objective reproducible
  alignment combined with VL inspection tools for
  SP1.6
• Method can also be applied to multi-spectral data
  cubes from other types of microscopes/telescopes.
• Data-cube256x256x65K. 100 cubes. Alignment15min
  in Matlab. Combinations (100 2) 15

13
Problem Meaningful ion dynamics

• Ion clouds 50k ions x 1M steps
• Current visualizations are low level, eg.
• But how about
• Intra ion-cluster interactions and their causes
• Intra ion-cluster interactions?

14
Our novel approach

• Idea simplify images with
• Statistical parameterized icons
• Semantic camera control
• Approach
• Parameterized comet-icons
• Camera motion relative to comet dynamics

15
Example icons

• Ions groups
• Statistical ion properties of group
• Ion density dynamics

16
Example camera control

• Trapping motion
• Relative cyclotron frequency
• Tracks of Frenet frames

17
Impact ? Generic ? GRID?

• Improvement of mass accuracy understanding/control
  leads to enhanced protein ID in proteomics
• Software framework is targeted towards particle
  visualization. Semantics of icons/cameras can be
  added/changed/enhanced
• Near-future optimization of simulation initial
  conditions

18
Final SP 2.3 comments

• SP 2.3 is well on track
• Projected output
• GRID enabled toolbox SP2 layer
• Applications using toolbox SP1 layer
• However visualization PhDs are not mass
  spectrometry scientists!