Hacker News new | past | comments | ask | show | jobs | submit login
CT Scan of a Pumpkin (homestead.com)
491 points by zdw on Oct 31, 2021 | hide | past | favorite | 79 comments



If you download the video and extract the frames (like, with ffmpeg), and load it into 3d Slicer (slicer.org), you can actually get a 3D reconstruction out of it! Out of a youtube video of the slices!

https://i.imgur.com/wmPQvfn.jpg


I work with large-scale neuroscience imaging, and this is exactly how we compress 3D image stacks (i.e. 3d volumes) captured with confocal microscopes. Since adjacent frames are usually quite similar, there's a ton of redundancy that H.265 can exploit, and the compression ratios are amazing. For multi-channel volumetric imaging, we use ffmpeg to encode each channel as a movie and then combine all the channels into a single HDF5 file.


Oh wow, this is a great idea. How do you deal with the lossy compression? There must be a lossless codec which uses the redundancy better than deflate?


This format is meant for visualization in 3d, and even though it's lossy, it's "visually lossless" for humans. We to start with the archived lossless stacks (compressed with bz2) for any reprocessing.

What you're suggesting with a lossless movie codec would be a great addition, we just haven't had the need for it yet.


Have you tried any of the open source AVIF image encoding tools, derived from AV1 video encoders?


Do you get 3D compression artifacts?

I love the idea that you rotate a spacial dimension into a time dimension, and after decompression you get artifacts from the time dimension visualised in a space dimension again.


We don't get visible artifacts because we dialed-in the CRF value to our data. But that sounds like a neat experiment to try.


FIJI plugin?



Ah, I should have guessed, I've done my time on the farm


I wonder if this would work for my hospital CT scans I've had in the last ~5 years. They all come on a CD, with a software program loaded onto it with the scans to view it in. It would be cool to be able to 3d visualize it all


Don’t go through that much trouble! There’s an easier solution.

You can download free radiology viewers RadiAnt (windows compatible) or Osirix (Mac compatible). Your imaging is in DICOM format probably and you can use Radiant to export all of your slices into .jpg if you want. You can also do 3D reconstructions of soft tissue, bone, lung, etc.


>Don’t go through that much trouble!

You realize this is HN where readers pride themselves on the trouble that can be accomplished in a weekend.

>There’s an easier solution.

But what else is one to do at the weekend?


Try to turn it into an stl file, slice it and print it :-)


Do this with a couple of "perfect" pumpkin shapes. Create a way to 3D print these as a mold to make your own "pumpkins" to be carved without all of the mess and able to last longer through the season. No more petroleum jelly, no more soaking them in the tub.


>No more petroleum jelly, no more soaking them in the tub.

What. I feel like I am missing out on a whole field of pumpkin science here that I was unaware of.


It's a plant. Think of it as a cut flower. After cutting open a pumpkin, they are obviously no longer sealed. They start to dry out. You can rub petroleum jelly all over the carved sides to help slow down the drying out. You can also soak them in the tub, and they will pull in some of the water to help them come back into shape.

There's all sort of things you can do to prolong things once they've been cut/carved/etc. My mom was a florist and designer. I've been in/around productions requiring things to be preserved so items can be kept around as long as possible. You just kind of pick up a thing or two


There are also various chemical solutions that you can put on the pumpkin to make it last longer. However those are not recommended if wildlife is around. Moose, in particular, love pumpkins. I suspect deer are the same.


Sure the deer, but I don't want that stuff around me. Better living through chemistry is something I'm trying to avoid now that I can make my own choices about what goes in, on, or around me.


personally I'm looking forward to "this gourd does not exist"


The program in question (3D Slicer) is also one of those easier solutions. It can load DICOM files directly.


Horos (Mac) is another - I think it’s the freeware version of Osirix.

You can find InteleViewer if you hunt about as well (Mac and PC).


You can get aeskulap on Linux. It's not anything that would be usable by a radiologist, but I've used it on occasion to look at some images.


We have been working on an open source tech that works in the browser and render 3D CTs and MRIs without installing any software, check it out:

https://openview.health


fyi you've misspelled "HIPAA" on that page


That is really nice. Thanks for sharing!


Cool, thanks for the link.


When my wife when through invasive breast cancer some years ago¹, I got started with RadiAnt because it's easy to construct a nice looking3d reconstruction, but I quickly hit a ceiling what I was able to accomplish.

With slicer I was able to produced compelling pictures to highlight the objects of interest by using custom color scales and transparency. For fun I also followed some tutorial on bone segregation for 3d printing, something that is clearly beyond the reach of RadiAnt.

Still, I would recommend to learn the basics in RadiAnt by trying to see what is in the written radiologist report². You will have to learn quite a bit of arcane terminology but I found that process quite rewarding and strangely empowering.

1- 3 years after her last treatment she is still cancer free, I hope it stays that way...

2- Ask for a copy of the report when you get the DICOM DVD. If the scan was taken at a hospital they will probably redirect you to the medical archives and it assuredly won't be ready when you get the DVD but they can mail it to you.


You're in luck, because (assuming the scans are in a compatible format), this is exactly what 3D Slicer was designed for.


there are many DICOM viewers available, ranging from the fairly horrible plugins for imageJ to very sophisticated things with maya or COMSOL or whatnot.

OSIRX is often very good, though not free.


Horos is the FOSS version of Osirix and based on the same original codebase, which lives on Github – c.f. https://horosproject.org/download-horos/


Sadly, it's MacOS-only. Would be nice to have something Qt or GTK-based.


There are many free DICOM viewer in the Ubuntu/Debian/Pop_OS repositories. I don't remember which one I used, but at least some of them allow to visit your organs in 3D :)


I recommend InVesalius. It’s in Flathub (https://flathub.org/apps/details/br.gov.cti.invesalius)


The pumpkin flowers have 5 petals, and the female flowers have 5 stigma, but the seeds inside grow in 6 columns. Which I find puzzling.

Google brings up https://threesixty360.wordpress.com/2009/10/31/the-mystery-o... which points out a 3-fold rotational symmetry in the pumpkin. And if you look, the 6 seed-columns are really 3 pairs of columns.

Five and three seem less puzzling I supposed, because Fibonacci. Still, I was expecting 5.


I'm confused too, but look at this photo in the article http://randomfootage.homestead.com/pumpkinTangerine1.jpg

It looks like there is an empty slot at 11-12 in the clock, and also an empty slot at 7 in the clock. Each one looks partially split in two, so I can imagine that they are the missing 2 parts of the structure of the flower.


I am considering whether to have a CT scan done of my heart. I am 35 yo and healthy, but have bad history of heart disease in my family. No current issues for me though. My doctor recommended I do the scan just to see if I have any plaque buildup, but is this not a lot of unnecessary radiation at a fairly young age? I hear it is like getting 1,000 x-rays done. Can anyone qualified chime in?


The FDA has a great page on the topic with citations to medical journals: https://www.fda.gov/radiation-emitting-products/medical-x-ra...

The short answer:

> If you combine the natural risk of a fatal cancer and the estimated risk from a 10 mSv CT scan, the total risk may increase from 400 chances in 2000 to 401 chances in 2000.

Your specific CT scan of the heart might be less than half of the 10mSv dose used in this example so the radiation risk is even lower.


> fatal cancer

Lot of non-fatal cancer out there, too.


If you are in US, one thing to consider is the massive over-prescription of medical tests in this country, due to legal reasons - avoid being sued for missing a problem.

You might want to consult with a UK doctor for example to see if they also recommend a CT scan in your situation.

https://www.newyorker.com/magazine/2015/05/11/overkill-atul-...


For plaque buildup they normally do a quick low resolution scan (to calculate an “agatson score”). I did not know by heart, but according to webmd.com it’s equivalent to 1 year background radiation. Pretty and high resolution images like the pumpkin above are possible but require more radiation. They should use the simplest scan that answers the doctor’s question.

A good question to ask is ‘why’: if no treatment decision depends on the outcome of the scan, the downsides can easily outweigh the benefits. A healthy lifestyle is advisable no matter what the scan says.

I am no doctor. I work on software for cardiologists.


Thank you


Have a read up on what the dose is and what this equates to. It’s low, and importantly, the alternative tests are pretty invasive.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4789203/


Better get some X-Rays than staying with unknown and untreated desease. This could help you : https://www.nuffieldhealth.com/article/radiation-exposure-fr...

CT scans are way more radiative than your classical X-rays but they are also way less radiative that anything that could be really dangerous.

You are way more prone to major heart issues if it’s in your family’s genes and untreated than to have any issue with a ct-scan.


Thanks, that does help


I am considering whether to have a CT scan done

CT is an absolutely amazing tool. The amount of health information produced is astounding.

I had a CT done to follow up on an adenoma. As part of the evaluation of that scan, the MD noted something else that was easily treated but that could have been very serious in not too distant future (I'm being vague for personal reasons).

Here are a few snippets from one of my reports:

Normal heart size. Clear lung bases. No pleural or pericardial effusion.

Remaining noncontrast liver, pancreas, spleen, and left adrenal glands appear normal aside from benign 1.7 x 2 cm right adrenal adenoma.

No abdominal aortic aneurysm with mild circumferential atherosclerotic calcifications.

So, e.g. the MD noted mild atherosclerosis as just one of his "incidental" findings. Good to know.



No pumpkin


Number 13 might count, depending on the definition of pumpkin.


No upvote


> Next, it's obvious how much air there is in a pumpkin. All the black on these source images represent air, quite a bit of its overall volume.

This made me really curious. The pumpkin is airtight and develops entirely within that airtight seal. It's full of gas pockets. How similar are the undisturbed contents of those gas pockets to atmospheric air?


A pumpkin is not airtight. Like most living things, plants breathe in O2 and breathe out CO2 (in addition to photosynthesis which does the opposite). So the air inside a pumpkin is most likely atmospheric air with less O2 and more CO2, maybe other gasses that participate in the pumpkin metabolism.


> Like most living things, plants breathe in O2 and breathe out CO2 (in addition to photosynthesis which does the opposite).

At what "level" does this happen? Is it similar to an animal with a diaphragm centralizing the intake of oxygen and lungs centralizing distribution to every part of the animal, or more along the lines of cells that are exposed to the air taking in oxygen locally and using it for their own needs?


Plants absorb oxygen, moisture, etc through their stoma found in the stems, roots, and leaves directly into their cells. From there its processed into ATP via the mitochondria just like we would.


'Most'? As opposed to all? I thought that was part of the definition, such as it is?


Many life forms on earth are anaerobic, powered by chemical reactions that do not involve O2. There are even species for which O2 is toxic. It most likely includes the very first lifeforms.

There are many definitions of life, some of them broad enough to include things like compute programs, there is a debate about viruses, but I don't know of any definition that requires breathing O2.


The one I'm recalling from school includes 'respiration' rather than O2 specifically. (Now that I've woken up.)


I wouldn't be surprised if gas exchange occurred through membranes, so that the gas inside is the same composition as the gas outside.


I'm pretty sure these pumpkins have been made to have more air for Halloween. Much easier to carve out.

Everyday edible pumpkins have much more flesh.


The most powerful CT Scan has been unveil very recently, with a magnet able to pull almost 12 teslas (where regular ones usually go to 1,5 to 3 teslas)[1]. Source is in french, but you'll find a similar video of... a pumpkin :)

They explain in the article that pumpkins are used because they share a similar shape and structure as well as water ratio as a human head. They got the magnet from previous research efforts from a research team dedicated to that domain (part of the french research around nuclear and alternative energies). Instead of trashing the 130t magnet, they proposed to adapt it for medical imagery. It will know help research on the Alzheimer disease.

https://www.francetvinfo.fr/sante/decouverte-scientifique/sc...

Edit: format


That's MRT - magnetic resonance imaging. CT scans - computed tomography - use x-rays.

https://en.wikipedia.org/wiki/CT_scan

https://en.wikipedia.org/wiki/Magnetic_resonance_imaging

The applications for both are slightly different.

https://www.medicalnewstoday.com/articles/154877#possible_ri...

There is a nice course on edX (the content is free), "Introduction to Biomedical Imaging", which teaches the basics of the most common imaging technologies (also adds ultra-sounds and radiology-based scans):

https://www.edx.org/course/introduction-to-biomedical-imagin...


Those were high resolution slices of CT scan and nice visualization as well. One can only hope human CT scans could achieve this level of preciseness without increasing the radiation dose.


> One can only hope human CT scans could achieve this level of preciseness without increasing the radiation dose.

It looks like a pretty standard scan to me and the article says the protocol used is an extremity protocol.

I work in radiology and various non-human things go though the scanner from time to time - fruit, pets, fossils, broken equipment that needs internal visualisation (sort of like the patients).


Well the article says that each slice is 600 microns apart so definitely smaller than a typical human CT that I’ve worked with where each slice is at least 1 mm apart. Also, the contrast is very high which means higher radiation does per slice is needed. In human context, like others have stated in this thread, if you don’t need a higher resolution then don’t order one. But sometimes, you only know if some areas needed higher resolution after you’ve seen the result. And the radiation dose for an additional, higher resolution and smaller areas might be much greater than just capturing the original CT with higher resolution.


> Well the article says that each slice is 600 microns apart so definitely smaller than a typical human CT that I’ve worked with where each slice is at least 1 mm apart.

I’ve just checked our PACS. The extremity protocols are at .6mm, face at .75mm and spine is 1mm.

It was .6mm 10 years ago when I ran them at other sites, so I don’t think this is anything pioneering.

You possibly do the same, but we generally reconstruct thicker. 1mm acquisitions reconstructed in another plane at 1mm look a bit gross as the acquisition needs to be thinner than the recon.

The dose is not high on modern equipment.


Sorry to be such a bore but when scanning humans the images should only be good enough to make a good diagnosis. If they are better than needed, the wise thing to do is lower the radiation exposure.


Roughly how much higher is the dose required for this level of detail than that which is considered safe for humans?


As far as I can tell, it’s not higher dose.

> It was scanned on a GE Revolution CT machine. The scan was performed using technique optimized for human extremities, and employed very thin slice thickness of 600 microns.


No level of radiation exposure is safe. How much of CT radiation is considered acceptable depends on the application.


Dose response being linear no threshold is the operating assumption. Radiation hormesis may be a thing but it’s difficult to be certain at low doses.

https://en.m.wikipedia.org/wiki/Radiation_hormesis


We build an ideal 3D Volumetric Viz tool: https://tomviz.org/ . Open-source and built for reproducible open-science. It focuses on volumetric data analysis, not just surfaces. Hopefully, HN finds it useful.


Is “They did a CT scan on a pumpkin” the next evolution of “They did surgery on a grape”[0]?

[0] https://knowyourmeme.com/memes/they-did-surgery-on-a-grape


Maybe it's the next evolution of doing an fMRI on a dead salmon: https://blogs.scientificamerican.com/scicurious-brain/ignobe...


As a partially texture-based eater, looking at these images makes me feel triggered.

But it also makes me feel grateful for food processors, blenders, etc. that can puree pumpkin into delicious pie filling.


Happy Halloween!


It certainly triggers my trypophobia!


Homestead… now that’s a host I’ve not seen in a long time. Good example of websites as they used to be.


I came here to comment this! So many old game fan websites still link to a bunch of broken Homestead sites. I’m amazed they’re still around. Got a little nostalgia when I saw the domain.


Well that's one of the more interesting things I think I'll see today! Neat.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: