Look Inside with CT!

A couple of weeks ago, <a href=“https://lookinsidewithct.tumblr.com/post/165232374232/shall-we-play-a-game-called-identify-the-ct-cross>I posted a mystery object</a>.

Behold! That same object in a cross section slice gif, from a different perspective!

And yet another perspective!

Yup, acorn. I tested a bunch of people on the initial video--folks here, and also friends and random people who stopped by the lab for tours. Most people guessed some kind of plant with rotational symmetry, which I thought was an interesting common thread. The suggestion of okra particularly delighted me, as okra pods have little caps too, and I hadn’t considered how similar they are to acorns before that.

In any case, playing this game got me thinking about one key bit of learning how to do scientific communication: making images that say what you want them to say. I’m in no way an expert, but since I’ve started doing CT, it’s something I’ve been giving a lot of thought to.

If I’m making an image, the question I have to continuously ask myself is: “What am I trying to get people to understand by looking at this?” Over here, I’m perfectly happy to show out-of-context mouse farts or a cross section of a fish that happened to remind me of an Allie Brosh cartoon, but if I’m trying to explain a real scientific concept to someone else, I need to make sure they know where they are in space, and what they’re looking at.

I’m giving a talk next week for which I needed to make a bunch of new good-quality images, and have been spending a lot of time with this. One thing I do a fair amount is just what I did here--grab a friend, ask them what the image looks like to them. Most of the images I was making for the talk worked well, but one particular one I struggled with because I was trying to put too much information in. The colleague I showed it to didn’t like it. I cropped it down to only the relevant bit, and suddenly it was very obvious to them. So, I needed two images. In the first I showed the stripped-down bit, and only after that could I add the context of where the interesting part was located.

Since some of the sciblr folks are early-career or aspiring STEM researchers, I thought I’d talk about the place I work, and explain that little piece of How Science Gets Done. As someone who came to this from a working-class non-technical family, I didn’t know how any of this stuff happened or who to talk to, and I missed out on tons of resources as a result.

So! I work in a core facility, which is one type of lab where research can be done. Labs come in a few different types.

-There are classroom labs, designed for the maximum number of students to get hands-on experience at once. It’s not common for them to be used for actual independent research, unless a professor has structured her class in a way that produces some data, or has some unusual-for-her project that requires really common equipment that she doesn’t normally need.

-There are labs that belong to one, or a very small number, of professors. If you do an independent research project or “join a lab”, this is probably where you’ll spend most of your time. The prof has purchased equipment, chemicals, etc, for that lab, and he’s probably carefully spent that money on things that align closely with his projects, that his students will use frequently, and that he wants a lot of control over. But there are downsides to buying stuff for your own lab. It’s like buying books (I’m talking experimental research, here, so assume paper books for the analogy to hold). Books look nice, and are nice to learn from, but they take up space, and they cost money, and their resale value is minimal. And if you let your students loose among them, they may spill some toluene in them and put them back and never tell you until you use them again and ARGH anyways.

But what if you need that equipment once a year and it takes up a ton of space? What if it’s a new shiny toy and you don’t have a million dollars to blow on it, just to find out it’s not that useful? That’s where core facilities come in.

-Core facilities are shared labs, a little like a library for instrumentation. Someone else maintains the equipment (hopefully), and either trains all comers on the equipment, or takes in samples and sends you data. Unlike a library, usually need to pay--either by the hour, or by the sample, or by the day of usage. The charges cover some of the costs to run the instruments, pay the staff, buy consumable supplies, etc, but it usually doesn’t cover all of it. The rest is covered by grants*, gifts, or professors chipping in to pay extra for something that’s important to them, but not enough to pay for *all* of. Core facilities usually have really expensive equipment, and staff that are very knowledgeable about operating the equipment.

Core facility staff, like me? Our whole job is to help researchers. We usually expect that you are not the expert on the equipment you’re about to use. We usually don’t expect you to know the first thing about it! It’s okay to say “My prof asked me to use this. Why?” or “Hey, I think this would answer my research question, can you help me figure out if that’s the case?” Meeting with students to talk about their research project is about 20% of what I do, and I always feel bad when someone has clearly been floundering for a while before contacting me.

We’re not here to do all your work FOR you. Sometimes I do have researchers who expect us to run their samples for them (fine, for the stated per-hour cost), and do some analysis (fine, for an additional cost), and research their specific project or write a report of the findings (not fine, especially for students). But that’s rare. Mostly our job is to help, and we probably know more about the techniques we specialize in than your prof, and have more free time too. Ask us stuff! We like to talk. As this wall of text indicates.

Core Facility Advertisement complete. Want to see what resources your university has? Search for core facilities and see what’s there. Different universities arrange their cores differently; some are all under one umbrella but most operate pretty independently. Schools focused on undergrad education will probably have fewer resources on this front, but even they are likely to have some shared instrumentation funded by individual departments.

One of my machines has been acting super flaky for a few months; basically, the x-ray light source was not giving a constant light but was flickering brighter and dimmer. This causes weird artifacts in the final dataset, little streaks formed as the software tries to figure out what all those brighter and darker images mean when combined. The company technician was able to log in remotely and restabilize the beam, but that restabilization was lasting for less and less time until it wasn’t even lasting overnight.

Finally they agreed it was time to get a new light source.

How many people does it take to change a light bulb* on a million dollar piece of equipment?

Only one, but it does take them a week.

But it’s back, and in good shape! Which is great news because I have a backlog of samples that has been building up while I do as much as I can on the two functioning machines. Having the third back and reliable will be hugely helpful.

Here’s a cool application for CT: non-destructive 3D analysis of archaeological artifacts! I’ll do one example this week and another next week.

Someone brought in some cuneiform tablets and wanted to see inside. Apparently some of the clay tablets that archaeologists have found are a bit like an envelope. Some writing would be done on a piece of wax, then the wax would be wrapped in clay, which could also be written on. At the time, breaking it open to read the message would have been no big deal, but nowadays researchers would like to avoid destroying rare artifacts just to see what’s inside.

Enter me, he says in parentheses.

The whole tablet is pretty small, a size that could be easily carried around in your hand on the streets of ancient Sumeria. It’s a little smaller than a packet of Post-Its. Here is the reconstructed CT data showing one face of the tablet:

(Don’t ask me what it says, or even if the writing is the right way up. I just take the pictures.)

Even looking at the data, you can see how fragile this tablet is. There are cracks and a few chips on the surface. but we can digitally slice it open to see what’s inside, and leave the artifact in the exact condition it started in:

And the same view, sliced down the middle to see inside:

When they pressed the clay into the written-on wax, the clay went into all the topographical detail, and we can see the original marks pretty clearly.

What we are looking at is a note made with human hands thousands of years ago and not seen since. Neat, huh?

Anyway, the impressions in the clay will be inside-out and backwards, and that might be hard for someone to read. Fortunately, I have software that can fill in the empty spaces and reverse that reversal. The animation below shows what that reconstruction looks like in the context of the original object.

And here is what it looks like on its own. The cracks are visible here but could be cleaned up with some time and care. Again, this blue object is just the filled-in inside of the clay tablet, and would not normally be visible without destroying the object.

CT images of a seedpod from an allamanda plant. These scans are part of an art project that a student is doing; they gave me permission to use the data and make my own images with it. The overall seedpod was about 2cm including the spikes, for scale.

I haven’t shown a ton with animation or image manipulation here before; this data happens to be fun to play with. Once you have the raw data, there are a lot of things that the software lets you do. You can select one set of brightnesses from the gray scale data and get a solid object:

So spiky, much points

Or you can leave the grayscale data as-is and slice through it to digitally dissect it:

Two little flat seeds inside!

Here’s what the inside looks like if you digitally remove the seeds and crop enough of the pod to see the inner surface:

Cross-sectional thin slices look a bit like a child’s drawing of the sun:

INFINITELY ROTATING THICK 3D SLICE