Jump to content

Dr. Mike

  • Content Count

  • Joined

  • Last visited

  • Days Won


Everything posted by Dr. Mike

  1. You are diving into the deep topic of medical imaging scans for 3D printing. I wrote a blog article on how to get the most out of your scan here. Take a look as it is very helpful. 3mm is not bad for a CT scan for 3D printing. In order to understand why the radiology department created 3 mm slices, you need to understand how CT scans work. Modern CT scanners do not acquire data in slices. Rather, the acquisition is helical. The patient moves in the scanner on the Z axis at a fixed speed, while the x-ray tube spins around that axis during acquisition. As a result, relative to the body, the x-ray tube moves around like a helix, i.e. corkscrew. The raw data from this acquisition is stored in memory in the CT scanner. It is then reconstructed into flat slices that can be of any thickness anywhere from 0.5 mm to 5 mm. Thinner slices are not always better however. There is a fixed number of x-ray photons that were acquired during the scan. When the slices are created after the fact, the data that those photons created is spread among the slices. If you have a lot of very thin slices then there are few photons per slice. Just like with a handheld camera when you shoot in low light, having a low number of photons results in a grainy image. The exact same things happen with a CT scanner. Thin slices tend to be very grainy and it can be difficult to detect abnormalities when the image quality is poor. Thicker slices on the other hand are fewer in number and have more photons per slice and thus are less grainy. Think of a nice photograph from a handheld camera on a bright sunny day. There are so many photons to make the image, the image is crystal clear. Thicker slices, while giving a higher quality image, are also thicker, and very small structures are harder to see. Therefore, if you had a CT scan and reconstructed 0.5 mm slices, those images would be much grainier and of lower quality than if you had reconstructed with 5 mm slices. When the scan is being taken the radiologist determines what slice thickness is the best for diagnosing the problem at hand. If you're getting a CT scan of the abdomen for appendicitis, you will get 5 mm slices. If you have a problem with your middle tier and are getting a CT scan of the mastoid, you'll probably get 0.5 mm slices. It should be noted that the raw data from the scan acquisition takes up a lot of memory. While the reconstructed slices are saved in the hospitals radiology system, the raw data from the scan is stored on the physical scanner itself. Typically after a few days that data is purged to make room for new scan data. So, after a few days it is not possible to generate any new slices from the scan, as the raw data has been purged. My guess is that 3 mm slices is as good as you are ever going to be able to get from your scan. To reduce the stairstep artifact, run a smoothing algorithm on your model. This should reduce that appearance. I hope this helps Dr. Mike
  2. 1) Nothing in this forum should be considered medical advice. 2) The scan shows some extent of pectus excavatum. Heart is a bit displaced as a result. 3) Evaluation of the heart itself is poor because the scan was not protocoled to examine the heart (i.e. no ECG gating). My guess is they were looking for PE, and there is no obvious one. Good luck
  3. If you increase the threshold value (150->250 Hounsfield units), that will tend to include less bone in the model. If decrease it, more bone will be included. Just want to check -- are you starting with a CT or an MRI. CTs work better. Hope this helps. Dr. Mike
  4. Dr. Mike

    OsiriX tutorials

    I use 3D slicer and the segmentation module. It takes a little time to get familiar with the tools, but they can be pretty powerful. I just did this kidney yesterday, including the kidney tissue, renal collecting system, artery and vein. FYI, we are building the ability to automatically segment organs into democratiz3D. Right now it only supports creation of bone models, but in the future auto segmentation of organs will be a feature. Hope this helps, Dr. Mike
  5. Try this tutorial. Creating the nrrd is pretty simple and takes about a minute. If you are trying to create a bone model from a CT scan, try democratized -- embodi3D's cloud service for automatic conversion. If you want to do other things, we have a variety of free tutorials online. Good luck!
  6. I think flaviu knows what a medical scan is. He is asking for specifics. CT? MRI? with contrast? If MRI what pulse sequence? T1? T2? MRAGE? post con? What body part? What slice thickness? etc etc. Hil, I think you may be underestimating the difficulty in printing from a medical scan. How are you going to extract your model data from the medical scan? How are you going to quality control the segmentation? How are you going to print it (size, material, layer thickness, orientation, supports, etc). You have to have this all planned out before you go down this road.
  7. Devarsh, your article is spot on. MRI is incredibly valuable, but there are huge limitations to 3D printing from it. Usually, I am stuck with doing a lot of manual segmentation, which is extremely time consuming. Thanks for writing this great article!
  8. I've never heard of this. It almost sounds too good to be true. Anybody have experience with this?
  9. Dr. Mike


    I think you can use 3D slicer. Here is a forum discussion on the topic. https://discourse.slicer.org/t/open-tiff-files/1494
  10. You are very welcome. We really try to listen to all member feedback, and appreciate it when people take the time to let us know about a problem. Thank you.
  11. Agreed. This limit was put in place to prevent abuse for free downloads, but should not apply to paid items. We are working on a fix. You should now be able to download your files. Thank you for letting us know about this issue.
  12. You can use a variety of free tools, including Blender or Meshmixer. Meshmixer is probably easiest. Here is the function shown in this picture.
  13. Kopachini is correct. You need to close the open hole.
  14. Version 1.0.0


    This 3D printable model of a normal human heart was generated from an ECG-gated contrast enhanced coronary CT scan. The slices are cut to illustrate the echocardiographic short-axis view. If you are interested in a 3D printable heart that shows slices in the anatomical transverse plane, click here. Notches have been added to ensure the slices fit together and do not slide against each other. The model demonstrates the detailed anatomy of the human heart in exquisite detail. Each slice stacks on top of the prior slice to form a complete human heart. There is a sturdy and stable base. Individual slices show the detailed cardiac anatomy of the right and left ventricles, and right and left atria, and outflow tracts in echocardiographic short-axis projection. Perfect for educational purposes. Technical parameters: Individual STLs: 4 Mesh integrity: manifold STL (watertight) vertices: appx 550k per object triangles: appx 1.1M per object dimensions: 11.4 x 9.1 x 10.6 cm


  15. We recently 3D printed a multimaterial skull with MCA aneurysm from a CTA head for customer who needed the skull in rigid plastic and the vessels and aneurysm in flexible material. The model will be used by neurosurgeons to practice intracranial aneurysm clipping surgery. To properly simulate the surgery, the skull needs to be hard and the vessels elastic. Combining two materials (and two printers!) provides the best solution. The model was created on democratiz3D. You can learn more about embodi3D's printing service here.
  16. You might want to check the spine and pelvis download area. https://www.embodi3d.com/files/category/13-spine-and-pelvis/
  17. There is a Medical 3D printing conference in Germany this May. Materialise has a conference in Belgium this June. There is a smaller conference run in Basel Switzerland. Here is the link from last year (none available this year yet) I am still looking for North American conferences. Will post if I find any.
  18. No, unfortunately it was cancelled for this year. I think the 3D printing conferences and workshops are going to organized by the RSNA 3D Printing Special Interest Group going forward, but I am not sure if they have determined an exact time and place yet. I highly encourage you to attend these conferences. They are highly informative and all the major groups doing medical 3D printing work (Mayo, Ottawa, Walter Reed, Stanford, VA, etc) will be attending and probably so will I! I will inquire with the RSNA 3D Printing SIG about the schedule of upcoming meetings and post a response here. Dr. Mike
  19. You need volumetric (3D) data, not 2D angiographic data. Unless you have a CT or MRI, angios wont work. You could potentially do a cone-beam CTA in the angio suite if you have the right software though. kopachini, thanks for helping out this member with your response. Dr. Mike
  20. I have extensively used the Form 2 as well as the Ultimaker 3E and other FDM printers. My review of the UM3E is below. In short, neither works well for the complex geometries in medical and anatomic models. The build volume for the Form 2 is a major issue unless you are only doing small parts like dental. Also, cleanup is a pain with support removal, resin cleaning and post curing. You will also run into problems like being able to remove the part from the build platform. Because the part is attached so tightly, often a spatula doesn't work nor does the Formlabs part removal tool. I cut an artery in my finger once and sprayed blood all over the wall because a razor blade is often the only tool that can separate the part from the platform. You will need to wear gloves for everything, and I know of at least two people who have developed skin allergies to the resin. The UM3E advertises a large build volume for anatomic parts, but in my extensive testing I ran into about an 80% failure rate. You can see my full review of the Ultimaker 3E at the link below. I encountered all kinds of failures, especially with large parts, and I don't think the larger and more expensive S5 really addresses any of these issues except maybe having a filament detector. Really the only way I have found to reliable print anatomic part is to build highly customized printers and software. It took about a year of intense trial and error to get everything right. I took a really deep dive, including reprogramming firmware in Chinese (which I don't speak). If you are willing to go this route, I encourage and applaud you. FYI, if you have medical or anatomic prints and want to save yourself the headache, embodi3D has recently launched a medical 3D printing service that will print and ship to you. It is an alternative to consider. With regard to cost I think FDM is clearly less expensive. Formlabs was revolutionary when their first printer came out, but the FDM space has much more innovation with open architecture. It kind of reminds me of Apple vs PCs in the 1980s - well engineered but closed architecture vs cheaper open architecture. Hope this helps and thanks for being an embodi3D member! Dr. Mike
  21. A member had a question about how to postprocess this print. I posted my response in the forums.
  22. A member recently messaged me with a question about a brain she printed from this file. I as posting the response here in the hope that it will help others in the community. QUESTION "I came across your 3D printable human brain model and was able to successfully print it. Thank you for sharing it! Now I need to post-process it and am wondering if you can explain how you post-processed your print? I have never done the post-processing before and am not sure the best approach to take. I have attached a picture of our printed brain for your reference. Thank you in advance for insight you can offer!" RESPONSE: Based on the picture you attached, it looks like you used a single extruder printer and printed both the supports and model in the same material, presumably PLA. You need to tear off the supports using pliers. This can be a time-consuming job as getting in every nook and cranny can be difficult. If you find the supports are stuck to much to the model, you may have to adjust some of the settings in your slicer software to compensate. There may be a rough surface where the supports touch the model that you can sand off. If you have a dual extrusion printer, you can print the supports using a water soluble material such as PVA, which makes the supports easy to remove by soaking in water. Dual extruders can be finicky and you will likely have to spend a lot of time trying out different settings to get the supports to work just right, including calibrating the XY offset of the second extruder, determining optimal print temperature for the PLA and support to work together, overhang speed, support infill percentage, etc. This process is very time consuming but gratifying once you get your printer dialed in. If you don't want to deal with the headache, embodi3D has a 3D printing service and can print and ship to you. Hope this helps. Dr. Mike
  23. Interesting. Thanks for sharing.
  • Create New...