Applied Science - 2013-01-08
I built a CT scanner from an x-ray tube that I bought on eBay, a stepper motor, a large ring bearing, and an Arduino. I used a phosphor screen and my camera to capture x-ray shadow images of a frozen chicken, while the Arduino and stepper motor rotated the chicken by 8 degrees between shots. The resulting 45 images were combined via filtered back projection to create a 3D volume reconstruction of the chicken. Software used: Panasonic SilkyPix (for .RW2 development) Adobe Bridge/Photoshop for image perspective correction Cygwin/Octave for filtered backprojection Cone Beam Computed Tomography algorithm http://www.mathworks.com/matlabcentral/fileexchange/35548-3d-cone-beam-ct-cbct-projection-backprojection-fdk-mlem-reconstruction-codes-for-matlab-students 3D Slicer for visualization and volume rendering http://www.slicer.org/ Axial image stack: https://drive.google.com/file/d/0B4YXWiqYWB99Z3pTRUE5WDZrdVU/
I work for Valve (video games) in their new hardware lab. When I am not at work, I spend most of my time on DIY/tech/science projects for fun.
You work for Valve huh? So when's Half-life 3 coming out?
@@jasonmoore1900 next epoch
Do you still work for Valve? Love the x-ray vids!
You blow my mind. Every time!
RIP
get rekt
@@Some_Beach dude he fucking died wtf
he blows you everytime? GHEY!
no i dont
Yes, a mirror would be a good way to get an undistorted view without placing the camera in the X-ray beam. I believe using a camera with sufficient pixel density and undistorting the image is easier than carefully positioning a large mirror.
Hi, please where can i buy your phosphor screen ?
I know I'm super late to the game, but I'm a radiology resident currently studying for boards. I'm learning all about x-ray production and CT physics. You sir are something else to demonstrate the basics of CT and even show how filtered back projection works to create a cross sectional image. Great work!!
Jeez...scanning electron microscopes, Xray machines, backscatter imagers...in my workshop the most I do is solder things and fix Commodore 64s.
This guy is over the top.
Got a real lab in the garage.
You are an American hero. I'm blown away by what you can do in your shop. Please keep making videos.
+Ryan Lucas couldn't agree more, people like this is why humanity evolves, a modern day Thomas Edison or Einstein right here. Imagine what could be learn if those earlier inventors and creators were able to share knowledge and document their ideas like we are doing now with youtube. Our creations and ideas must be shared and taught for the most potential to be attained from humanity. The entire arduino community is awesome and it's really set a precedent, i've seen many more companies offering open source projects than ever before lately. Love it.
Thanks! I still have the chicken in my freezer for more imaging experiments.
Still?!
I'm a service engineer for Siemens CTs, And I'm really impressed with the method you're using. I recommend you to keep developing this method because from the detector side of Siemens and other brands CTs are using way too much complicated detectors with lots and lots of complicated timing between readings and reconstruction of images and a lot of image artifacts making them very expensive. Loved your work and I'm pleased to help you for any hints or maybe parts from Siemens CTs.
It's amazing the ionization of all of those xrays didn't error out any of the electronics in the camera or arduino.
You are correct about the mirror. I think some commercial fluoroscope systems work this way. I used Cygwin/Octave with someone else's code for backprojection. Then I used 3D slicer for visualization. I put some links in the description.
I just found this video and must compliment you on your logical approach to solving the various issues. I worked for Siemens Medical systems for many years and also Mallinckrodt Corp. in their medical X-ray division. I have more lately moved into dental X-ray digital imaging where a CMOS sensor replaces the traditional films used to take X-rays. I have worked on all of the imaging modalities to different degrees, my favorite being cardiac imaging. Your project is great and the item you need to process those images is an image processor like those used in the big CT machines. Also your X-ray source is much smaller but about right sized for your project. All these items are large and super expensive but you are on the correct path following what the big machines basically do.
I recently purchased 2004 ZBV backscatter. Anybody want to help me sale it and make money together? Before I decide to strip it down and turn it into a food truck. I heard its about 750K brand new. Xray units still look like brand new. Please email me at saipandivermaster@gmail.com and my name is Patrick. TIA
My mother died of brain cancer. If all oncologists had a compact easy to use scanner that can scan only a patients head she may have had been able to get treatment before having symptoms. It could also help GPs make quick assessments of head injuries without the wait to go to a full size scanner. Please continue developing this more compact scanner. You may be able to save lifes through early detection of cancer. Thanks
As an engineer/MD/Comp Sci/PhD I'm amazed at what you did here! I just want to say hats off as this is incredible work! In spite of my myriad of degrees and educational experience I would say that this is something I am likely incapable of putting all together except for a lifetime of work!
I would recommend that you consider using Android for this type of automation. The ADK offers the ability to connect an Android device to custom electronics, and a very rich programmable pipeline. I'm sure you could have automated most of the camera control, capture, and photo manipulation (eg. imagemagik) via one device, and a handful of scripts. This could offer faster and more flexible development! Again, outstanding work!
Amazing stuff! The amount of hard work and knowledge distilled into a mere 6 minutes boggles my head. I doff my cap Sir.
Jesus Christ... That's amazing. I'd have been blown away just by you getting a short movie of the rotating chicken taken from the frames of the exposed phosphor sheet. That you then processed it in post using on hand off-the-shelf software just takes it to a whole other level. The YouTube "like" button is not enough.
This is very, very impressive. In fact, the model of vertical rotation that you have demonstrated would probably drastically reduce the cost of traditional CT scanners for patients that are capable of standing up (I'm guessing many scenarios). A static exposure plate may yield superior captures as well! Great work!
Current CT scanners use RF or light to transmit data. Toshiba scanners transmit data using a series of LEDs called mudat. They all transmit power by brushes and slip rings. Scanners I'm familiar with transmit 550 volts DC across the slip rings. Scanners from the 90s used slip rings to transmit data as well as power. Scanners in the 80s and older used cable take up systems. The scanner would make one rotation and take one slice and then it would have to stop and wait to rotate back before it could take another slice. I don't think scanners that old did 3D.
Amazing setup. I work with CT scanners, but I wasn't searching for CT when I found this.
kel5944: The 1980's ones sound interesting. I bet they could easily be updated to get many slices and then you can make a 3D texture using today's software.
Vrej Egon Spengler Today they do .5mm slices and 256 slices per rotation. 1 rotation in .5 seconds. There is a scanner that can scan in 1/120 of a second but is seldom used today and expensive to maintain. Its also not really needed either. The gold standard is to stop the motion of the heart in one rotation.
The 3d scans are awesome! You can even do 3d prints from them if you can get your scan data from your doctor. He wont spend the time to print it though. A 3d print isnt really useful to a diagnosis.
Vrej Egon Spengler There used to be stand alone consoles that you could input the collected data to create the 3D reconstructions
Except in bone fractures, then 3D is very useful
I've had a Linistepper for nearly a decade, using it for random motion tasks over the years. Aside from the curiosity value regarding how it accomplishes microstepping, it just works with any (unipolar) stepper I throw at it. Love that design, along with every last thing Roman Black has done.
You're a fucking legend.
I respect your technical ability Mr Krasnow! Wish I could do even a fraction of what you're doing.
you are a man of many disciplines
Great project 👏🏻 the filtered back projection explanation was a nice addition!
This sounds like an awesome application of vector mathematics :D
So what's the prognosis? Is the chicken going to make it? I hope it doesn't develop cancer later on and sue for malpractice. LOL
It's not good, the patient has a sever case of chicken butt.
@@simple_fred that must be the reason why the patient chickened out before prognosis ^^
Six and a half minutes and I now understand CT scanning. That's why I love this channel.
I like it… I see that I will have to study more, the difference between CT, CAT, MRI & X-Ray to get a clearer understanding of what you've done here. But, until then… Thank you for helping to create a more open and capable future for us all.
Holy crap. I had no idea a DIY at-home X-Ray could have detail anywhere near this good. I've actually wanted to try something like this for quite some time.
Yes. There is a 3D image matrix in octave, which can be exported as axial slices, then imported into 3D Slicer.
Thanks! I used Cygwin/Octave with someone else's code (link in the description) to perform the backprojection. I am currently working on an iterative reconstruction method, and will post more video of it soon. The graininess is partially caused by the backprojection algorithm creating artifacts, and also caused by the input images having poor dynamic range.
Awesome. How did you manage to reconstruct a scalar volume (axial slices) out of 2D planar projections? A tomography usually employs 1-dimensional parallel beam projections under a specific angle and then organized into a sinogram. So the back projection algorithm can approximate the slices out of such sinogram. But being a sequence of 2D planar X-ray roll (or a set of pics) is somewhat equated to a photogrammetry problem. Therefore, I figure out that what you did is a cone beam projection beforehand. But, I think that the provided matlab code only reconstructs previously calculated cone beam projections, but not the cone beam projection itself. Thx
Gotta love Arduino, and the versatility it brings...
I bet archaeologists and paleontologists would love a setup like that for scanning bones and such. Good work!
You can get radioisotope test sources from many suppliers. Do a google search for exempt radioisotope source and you can find them in many flavors. These are pretty much harmless, though you wouldnt want to sleep with one under yor pillow. They are used for calibration procedures in gamma spectroscopy, among other things.
That was awesome! Thank you for taking the time to create this and share it with us mere mortals. I'm in awe of you!
This is amazing. You've got a great production value on your videos as well as do an great job explaining all the steps you took. Thanks!
That's one of the most interesting things I've seen on youtube. Thanks.
now this is why we subscribe to your channel. enviable intellect and admirable passion kicking the worlds butt in ever single upload.
this is the best video I ever saw on YouTube about CT -great job
I appreciate people like you for teaching things like these... and I know that it takes a lot of time and effort to prepare such kind of videos... thank you very much. thumbs up :)
I like to watch your videos and act like I understand what your talking about.
No! People like Ben are needed on This Planet!
Me: "Hey did you see the game last night?" You: "No, I built a home made CAT scanner in my garage instead." Me: O_o
Man, you really are a genius... i might be able to reach your level one day, but sadly i live in a city... your workshop garage is the size of my house.
Yep, you're missing somthing else. I was talking about using the sensor strip from a flatbed scanner. It is basicly a single line of "pixels" that is moved on a track by a stepper motor (or a similar mechanism). I think they are just a bit shorter than that phosphor screen is wide; but with a custom track you could get much more coverage on the vertical axis.
What if instead of putting the camera in line, you use the old high speed trick and put it 90 degrees to the screen with a mirror set at 45 degrees to direct the light from the phosphor screen into it? Seems like that would be the best geometry to get straight on images while keeping the camera out of the X-ray beam.
R C because the phosphorous screen is affected by radiation, not photons.
HeATThER Mdd Phosphor screen produces light when struck by x-ray radiation, this is then reflected by the mirror into the camera. This avoids the camera being exposed to x-rays as this would produce hot pixels/noisy image.
fuck yeah, a mirror is invisible to xrays.
I came here to say this but u said it first.
He could put lead shielding glass in front of the camera and move everything perfectly behind the phosphor intensifier screen. She would certainly have gotten a better image and spent less time on Photoshop.
Awesome, congratulations to building a working CT scanner BTW: I work here: at the Tomography lab at the Paul Scherrer Institute in Switzerland, so I know what's involved :)
Wow. You sir, are my idol. I don't think I can ever acquire enough expertise in such a wide range of fields such as yourself and at the same time implement them in a practical and fascinating way.
yes yes i understand this, i actually asked implicitly why he chose 8 degrees because he didn't say why, i was wondering if it had something to do with the algorithm he used to create the 3d model, or if it's just limitations with the stepper motor, but you have a point, it must be just a compromise between, the quality of the final model and the time needed to render the whole thing :) .
OK just to try and plant some idears: DYI:MRI . I find MRI's quite more interessting. But even to scan a bug creating the magnet probably is quite a challenge. But hey before finding this channel I didn't even thougt about building things like a electron microscope.
@satchelfrost6531 - 2015-05-27
This guy is outlandishly versatile and better at everything than everyone.
@_I_Hear_You - 2017-11-25
this guy just have a lot of expensive equipment to play with them... and a lot of even better guys don't..
@fss1704 - 2018-06-21
+IandBee comp we fucking do our equipment.
@valterkaugust8511 - 2019-09-15
@@_I_Hear_You somebody jelly
@bigchungus1592 - 2020-09-22
@@valterkaugust8511 I'm not horizon, but I'm most definitely jelly myself. Not only for his intelligence and mechanical ability, but also his knack for acquiring expensive parts and gadgets.
@UncleKennysPlace - 2021-02-09
@@_I_Hear_You Yada. Yada, yada. I'm late, but you are wrong. Ben is outlandishly versatile. I say that as an outlandishly versatile person.