Tuesday, December 1, 2009

Data or Hypothesis?

Skull of modern human (Homo sapiens), reconstructed from CT scan of original specimen using 3D Slicer 3.4. Data from OUVC 10503, downloadable at the WitmerLab web page.

As the headline says: data or hypothesis? Discuss.

15 comments:

Mike Keesey said...

What's the hypothesis, that 3D data reconstructed from CT scans using certain algorithms is a reasonable proxy in many cases for the actual specimen?

Andy said...

That's one facet of the question, certainly. To rephrase the question a little more explicitly, is the image that I presented data or a hypothesis?

Mike Keesey said...

The image is data. Someone may have a hypothesis concerning the data in the image, but that hypothesis is not inherent in the image itself.

Tor Bertin said...

Any idea what the extraneous material around the orbits and nares is?

davidmaas said...

well... if its data, its abstracted data. Does that make it hypothesis?

Tor: those are artefacts due to closure algorithms used to make a polygonal mesh out of the point cloud. Either that, or there were some freak points in the scanning data.
Actually, the point cloud would likely be data and the poly mesh representation. Hypothesis needs a question for which it proposes an naswer, doesn't it?

Interestingly, I'm now working on a stegosaurus mesh (well, when I'm not doing my contract job) which is stylized so that it is clearly detached from the claim to be data. Here's my first test, a human male:
http://www.drip.de/?p=485

davidmaas said...

me agian:
Tor - I should extrapolate: the noise there might be a resolution artefact as well... ie. the details at that point surpass the resolution of the mesh, so the reconstruction kind of goes haywire. Mostly seen on high-detail structures like hair, or where the projection information was visually blocked or otherwise limited (corners, holes, etc.)

Casey said...

Its a hypothesis! simply the software's and operator's interpretation of the raw data, but the raw data is also just a hypothesis cooked up by GE when they built the CT scanner.

i think i know where you're going with this and won't wax on.

Larry Witmer said...

It's both a hypothesis *and* data, depending on context. It's a hypothesis to the same extent that a cast is a hypothesis in that both are representations of the original physical object, and so can be falsified by reference to the original. At the same time, both cast and 3D viz of CT data are data that can be used to test other hypotheses...at least I freakin' hope so...it's a big part of what we do. :)

As for the technical bits about "extraneous material," I actually have the original skull here in my office for reference. We have 3D viz work of this skull for comparison on YouTube (http://www.youtube.com/watch?v=3bCfBmOGJEw), as well as here: http://www.oucom.ohiou.edu/dbms-witmer/Human_movies.htm. The paper and raw data can be downloaded here: http://www.oucom.ohiou.edu/dbms-witmer/DinoSinuses_main.htm. The "extraneous material" is just an isosurface artifact caused by thresholding such that thin structures (e.g., inferior turbinate, lamina papyracea) are rendered incomplete. No slam on Andy intended here...he just posted a quick image to start the discussion.

Thomas R. Holtz, Jr. said...

New Shimmer is both a floor wax AND a dessert topping!

Andy said...

Awesome discussion so far! Let's keep it rolling!

(and no offense taken at all, Larry - the benefits/limitations of reconstruction vs. reality will be an important part of the upcoming discussion!)

220mya said...

Oh c'mon, its both a wave and a particle!

Anonymous said...

Data- information collected from observation

Hypothesis- educated guess based off prior knowledge of the subject.

One could argue it's data, but without any further information, it is, therefore neither data nor is it a hypothesis.

Unknown said...

Data. And, as any data, it must be treated with caution, having the limitations of data acquisition in mind.

If you measure by hand, you screw up, the caliper has a minimum accuracy, etc. Same is true here.

Alison said...

Classic. Classical science found it difficult to reconcile with the latest modern computers can offer.

John Scanlon FCD said...

All data is hypothesis: 'I think the caliper read 23.08 mm, so that's what I'll write down', and anyone else can come along later and check (if the material and method of measurement are specified precisely enough). Or 'what I drew, I saw' (but the eye and hand, let alone the dodgy office scanner and jpeg compression, introduce spurious details that become 'data' for the user). Error, noise and artefacts are always present in data, but that's what we build everything else on. It's OK!