Rapid Visual Inventory & Comparison of Complex 3D Structures

NSF Scientific Visualization Challenge 2011 1st place video:
Prototype of Brad Marsh's geometric abstraction concept to visually simplify complex data by Graham Johnson, Brad Marsh, and Andrew Noske.
http://www.wired.com/wiredscience/2012/02/science-visualizations-2011/?pid=30...

This entry demonstrates a new visualization standard we've developed so that complex three-dimensional (3D) datasets can be computationally 'morphed' into much simpler 3D geometric abstractions that are more "visually intuitive" than the original images. Our approach can be used to streamline the scientific analysis of multiple parameters simultaneously for complex 3D images and/or to more effectively communicate key insights afforded from such data to a wide variety of audiences (both scientists and the general public), including schoolchildren.

This animation explains how we've applied this unique process for rendering extraordinarily complex biological data (3D image maps generated by 'cellular tomography') into simplified geometric models that are much easier to interpret. The original 3D maps of pancreatic 'beta' cells precisely capture (at the 'nanoscale') both the extraordinary beauty and complexity of the molecular machinery within cells. Even for experts, however, such complex data remain difficult to visually interpret without painstaking image analysis. In contrast, the abstracted versions computed from the original 3D models employ simple geometric shapes to accurately depict the size, number and length of the various different 'compartments' or 'organelles' within each cell. In addition, these objects can be easily reorganized in 3D space, such that the simplified geometric models reveal "at-a-glance" how the complex internal landscape of insulin-secreting cell undergoes dramatic changes in response to different metabolic demands, physiological states and chronic disease conditions, such as diabetes [1,2].

1. Mapping the β-cell in 3D at the nanoscale using novel cellular electron tomography and computational approaches. Noske AB, Marsh BJ. In: BetaSys - Systems Biology of Regulated Exocytosis in Pancreatic β-Cells (Booß-Bavnbek B, Klösgen B, Larsen J, Pociot B, Renström E, Eds.). 2011. Springer. Series: Systems Biology, 2(8):147-183.
http://www.springer.com/new+%26+forthcoming+titles+(default)/book/978-1-4419-...

2. Expedited approaches to whole cell electron tomography and organelle mark-up in situ in high-pressure frozen pancreatic islets. Noske AB, Costin AJ, Morgan GP, Marsh BJ. Journal of Structural Biology. 2008. 161(3):298-313.
www.ncbi.nlm.nih.gov/pubmed/18069000

A figure [CELLuloid] that artistically summarizes this animation as a series of movie frames can be viewed at:
http://www.grahamj.com/bradmarsh/MarshCELLuloid/CELLuloid-Johnson_Noske_Marsh...


CONTRIBUTOR TEAM
Graham Johnson PhD: implementation of 3D geometric abstraction concept, computer graphic animation, narration
Molecular Graphics Laboratory
Department of Molecular Biology
The Scripps Research Institute
La Jolla, CA, USA

Andrew B Noske PhD: data acquisition, segmentation, image analysis
Structural Cell Biology Group
Division of Molecular Cell Biology
Institute for Molecular Bioscience
The University of Queensland
St Lucia, QLD, AUSTRALIA

Brad J Marsh PhD: development of geometric abstraction concept, data acquisition, image analysis, project supervision
Structural Cell Biology Group
Division of Molecular Cell Biology
Institute for Molecular Bioscience
The University of Queensland
St Lucia, QLD, AUSTRALIA

SPECIAL ACKNOWLEDGEMENTS
Mr Garry Morgan, The University of Queensland
Mr Adam Costin, The University of Queensland
Prof. Ronald A Milligan, The Scripps Research Institute
Prof. David N Mastronarde, University of Colorado

Category:

Science & Technology

Tags:

• Insulin
• DNA
• Diabetes
• Pancreatic Beta
• Bcell
• b NSF
• NSF Sci Vis
• SciVis
• SciVisChallenge
• Cinema 4D
• C4D
• ePMV
• MGL
• Scripps Research Institute
• TSRI
• Goodsell
• 3d
• dynamics
• morphing
• cell biology
• cells
• molecular
• molecules
• proteins
• watson
• crick
• illustration
• animation
• medical illustration
• medical art
• ami
• collision detection
• mograph
• bullet physics
• NSFcellsimplification1
• simplification
• morph
• mitochondria

[art4med]

Viewers would be aided in their understanding (wonderfully evident in your 3D pancreatic cell) IF you demonstrated WHY the "best fit" shape for most cells (or any group of plastic forms pushed together, even within a layer) is the HEXAGONal [flattening sides as pressures normalize]. Its one pet-peeve that most cells are represented as spherical (which would actually be rare)!

[grahamj21]

@art4med Fortunately, hexagonal shapes you suggest could be another option in the future as the algorithms for the shape morphing are fairly generic. While stacking everything as cuboids would show the total organelle volume more accurately, it is interesting in this movie that stacking the vesicles as spheres (the primitive closest to the native vesicle shape) gives a decent approximation of how densely the native vesiclescould pack... they are hexagonally packed in this version.

[grahamj21]

...Those shapes vary depending on the cells global position and number of neighbors in the cluster, e.g., youtube.com/watch?v=ipf1XE21rM­o

Thanks for the posts!

G

[grahamj21]

@art4med Lastly, as you can see in the original data, there are amazingly flat spots on the native cell (which was segmented from a mouse islet). Since these cells grow in 3D clusters and aren't restricted to polarized 2D interaction like tight/gap junctions as in endothelial and epithelial tissue, they have varied polygonal shapes making up their flat faces where, as you say, they squish up against neighboring cells....

[DJDepRemixes]

Sounds like Toby off of the office

[grahamj21]

@DJDepRemixes Good Call...never noticed that.

[DJDepRemixes]

@grahamj21 kinda of freaked me out with the similarity but great video.

[art4med]

Thanks for responding; of course I truly admire your work!

[art4med]

@grahamj21 Surely you are not calling for cell shape to vary as membranes like soap bubbles, the usual medium for visualization of complex math that is otherwise unsustainable? :' )

Whatever the case, you must admit that the spherical representation often seen in animation is more-often 'inaccurate' within "solid" media of tissues.

[art4med]

Congrats again on your award: I applaud your composition, color choices, and lighting as well. And thanks for the introduction (after some digging) to IMOD out of UCBoulder.

[grahamj21]

@art4med I agree, it would be great to change the shapes that objects morph to. This was our first pass of the algorithms and protocols that I created for Brad Marsh to prototype his concept ~2.5 years ago. There are many things I will change if I get time to go back in and develop it (hopefully to release it for others to use some day.) In this case, if we truly want to show total volume relationships, it would be wise to make everything cubic so the Y height truly corresp. to relative volume.