CT-Scanning the Archbishop
November 18, 2009
Last week, for the first time ever, I spent the entire working week on palaeo. I took a week away from my job, and spent it staying in London, working on the Archbishop at the Natural History Museum. (For those of you who have not been paying attention, the Archbishop is the informal name of the specimen NHM R5937, a brachiosaurid sauropod from the same Tendaguru area that produced Giraffatitan brancai, and which has been generally assumed to represent that species.)
My main goal was to take final publication-quality photographs that I can use in the description (which I have committed to try really, really hard to get submitted by the end of 2009). There’s quite a bit of material (more than for Xenoposeidon, anyway!) — six cervicals in various states of preservation/preparation, cervical ribs, two complete dorsals, two more dorsal centra and a dorsal spine, some scap scraps, a partial ?pubis, a long-bone fragment and “Lump Z“, whatever that is. What you see above is my best lateral-view photograph of what I’ve designated “Cervical U”. One of these days, I’m going to do a post on how to photograph large fossils — something it’s taken me five years to get the hang of — but for today, I want to tell you about an exciting adventure with Cervical U. [Update: I wrote the How To post a few months later.]
Because my other big goal on this trip was to get it CT-scanned. Thanks to the generosity of John Hutchinson of the Royal Veterinary College, and to the help of the NHM people in arranging a loan, everything was set up for my host Vince Bickers and me to ferry the specimen up to the RVC, scan it and return it.
But first it had to be packed:
Lorna and Sandra spent a long time looking for a crate big enough to pack the bone in, but came up empty — there was one that was long enough but not wide enough, one that was tall enough but not long enough, and so on. In the end we sat the bone, on its very solid plaster base, on a plastic pallet, and wrapped it in pillows, bubble-wrap and that blue stuff whose name I don’t know.
As it happened, the scan had to be delayed for a day due to lack of personnel at RVC, but Vince and I took the vertebra up on the Thursday anyway; he had to return to work on the Friday, but I took public transport to RVC for the big day. Before we went into the scanning room, John showed me his freezer room:
I found it amusing that they have enough Segments Of Awesome that they have to label the various elephant-part freezers differently. And further down the aisle:
Then it was off to the scanning facility, where we found that we had to unpack the vertebra: it was small enough to go through the machine, but there was no way the pallet was going through. Once we’d unpacked it and removed it, it fit pretty nicely:
Because the scanner spits out X-rays in all directions, it’s controlled from a separate room, behind lead-impregnated glass:
We ran three scans before we got the settings right — we needed more voltage to get through the bone and matrix than we’d first realised, and a filter was causing unhelpful moire patterns. The third scan was definitely the best, and the one I expect to be working with.
[Boring technical side-note: I plan to use 3D Slicer for visualisation thanks to Andy Farke’s series of tutorials. But, frustratingly, I wasn’t able to load the DICOM files from the scan into that program: it crashes when trying to load them (segmentation fault) even though it works fine on the ankylosaur skull that Andy walked us through in the tutorials. I fixed this by gluing the 300-odd files together into a single stack file that 3D Slicer was able to read. For the benefit of anyone else who needs to do this, the command (on a Ubuntu Linux box) was: medcon -f *.dcm -c dicom -stack3d -n -qc]
Here is an example slice, showing part of the condyle in posterior view:
The grey blobs at the bottom of the image are the plaster jacket that supports the vertebra; the white is bone, and the light grey inside it is matrix that fills the pneumatic spaces. I’m showing the condyle here because its cavities are clearly visible: further back in the vertebra, they are harder to pick out, perhaps in part because of the iron bars scattering the X-rays. It’s notable that this vertebra is less pneumatic than would be expected for a brachiosaurid — by eye, it looks like like the condyle is only 20-30% air, and this slice is not unrepresentative. Most neosauropods would be at least twice this pneumatic, so we may have an Archbishop autapomorphy here.
I’ve not yet persuaded 3D Slicer to build a 3D model for me, but I’m pleased to say that before I left RVC, John mocked up a quick-and-dirty render of the bone using only density threshholding, and I can at least show you that.
Here we see the bone from the left side, previously obscured by solid plaster. From a single static image, it’s not easy to make out details, but we can at least see that there is a solid ventral floor to the centrum … and that those two crossed iron bars obscure much that we would like to see. You will get more of an idea from the rotating video that this is screencapped from.
Looking at this and comparing it with the right-lateral photo at the top of the post, it’s apparent that the density threshhold was set too high when making this model: all the bone along the lower right margin of the middle part of the centrum is good, but it’s been omitted from the model. In other words, the vertebra is more complete than this proof-of-concept model suggests. Hopefully I will shortly be able to show you a better model.