Jump to content


Member With Blog
  • Content Count

  • Joined

  • Last visited

  • Days Won


ebaumel last won the day on November 29 2019

ebaumel had the most liked content!

About ebaumel

  • Rank
    Advanced Member & Blogger
  • Birthday 07/28/1959

Recent Profile Visitors

1,273 profile views
  1. Hi Mike! It would be great to meet up again at RSNA. I'm arriving in Chicago on Tuesday - leaving Saturday am. Eric Baumel
  2. Version 1.0.0


    CT_Shoulder_EMB_2_v0 - processed


  3. Vlad, Thanks for sharing the videos. It is very exciting to see this technology in the hands of clinicians.
  4. Every 3D printing case is different, and must be tailored for the individual patient’s specific clinical condition, anatomy, and imaging techniques. A 47 year old woman with a renal mass was being evaluated for surgical treatment planning. A urologist familiar with my current 3D printing work requested a 3D printed model of the kidney. The purpose was to help demonstrate the anatomy of the mass with respect to the renal hilum, to help determine if a partial nephrectomy was possible, or if a total nephrectomy was required. The patient had a documented reaction to radiographic iodinated contrast, and therefore an MRI was performed instead of a CT scan. The scan was performed on a 1.5 T GE Signa Excite system. The data set from a coronal 3D gradient echo pulse sequence acquisition was chosen because it best visualized the tumor encroachment into the renal sinus. The 3D model was created from segmentation of the kidney from the 3D gradient echo acquisition. The renal parenchyma was then made into the digital 3D model, leaving the mass as negative space. This was then printed to demonstrate the entire kidney. An additional 3D model was created showing a bisected view of the kidney along the coronal plane. This was done to see which model would be of more utility. The patient’s DICOM image files from the MRI were processed using the Materialise Innovation Suite’s Mimics and 3-matic software. Initially I used the Mimics software segmentation tool to segment the normal renal parenchymal tissue. This left a filling defect where the mass was. This negative space was useful for demonstrating the extent of the tumor. Using the 3-matic application I then took the 3D digital representation and created a model cut in the coronal plane. This helped better define the extent of the tumor invasion into the renal hilum. Both the full kidney and the coronal bisected models were printed for the surgeon and the patient to review. The STL (stereolithography or standard tesselation language) files generated for the 3D models were then imported into both the Cura and MakerBot slicing software applications to generate the gcode for the Ultimaker 2 and .x3g file for the MakerBot printers. Fused filament printing of the full kidney using Acrylonitrile Butadiene Styrene (ABS) on the MakerBot Replicator 2X Experimental Printer and sectioned kidney in Polylactic Acid (PLA) on the Ultimaker 2 printer. In general I prefer PLA to ABS. With PLA there is less shrinkage and warping of the material during the printing process. PLA is of plant based origin (here in the US it is derived from corn starch) and can print lower layer height and sharper printed corners. PLA, a biodegradable plastic is used in medical devices and surgical implants, as it possesses the ability to degrade into inoffensive lactic acid in the body. ABS is a petroleum-based recyclable non-biodegradable plastic. Unfortunately emits a potentially hazardous vapor during printing. The Replicator 2X however is optimized for ABS. http://pubs.acs.org/doi/pdf/10.1021/acs.est.5b04983 For more information about PLA/ABS see http://3dprintingforbeginners.com/filamentprimer/#sthash.p07fHBmh.dpuf The urologist showed the models to the patient, and it help to convince her of the necessity of a total nephrectomy, rather than a partial nephrectomy. Although the printed models showed the extent of the tumor invasion adequately for both the urologist and the patient to visualize, the models did not differentiate the mass from the renal sinus. To better demonstrate the tumor invasion, a new model with two different color filaments was created. Hand segmentation of the 3D model of the mass was performed due to the limited tissue contrast between the mass and the surrounding soft tissue structures. Creation of two separate segmentation files and 3D models was made in Mimics. This was then exported into 3-matic for local smoothing. Two separate STL files of both the mass and kidney were generated. The combined model was then bisected in the coronal plane and the additional two STL files were again generated. Once the combined STL files were imported into the MakerBot software, they were repositioned on the build plate in an orientation to optimize the printing process. The STL models were positioned on the bed, above a ring spacer file. This is necessary for the proper printing contact. The model was oriented to minimize the amount of printing support structures. When using the two filament colors, “purge walls” are generated by the software to help eliminate the small threads of filament from one color being deposited when the next color is laid down. The two filament model enables the surgeon and the patient to better visualize the extent of the tumor invasion, clearly demonstrating normal from abnormal tissues.
  5. Great post! I wish I had been able to read it before I started making my own mistakes. I look forward to the future articles.
  6. Very nice overview, Tatiana. I'd love to see a post on potential applications in radiation oncology treatment planning.
  7. I'm so gratified that you like the post, and the model! Regarding support material in the lumen: with these models it wasn't an issue. There was no internal supports except for a very thin support at the openings, which were easily removed in post processing. In future posts I'll try to show some of the physical model making steps.
  8. This is a time of rapid growth in medical 3D printing. The technology allows us to take an individual patient’s scan information and create physical models, which can be used in any number of clinical applications. The industry standard DICOM image files from CT and MRI scanners can be converted into 3D files, such as STL (for stereolithography) files. These digital models can then be uploaded to a 3D printing service bureau or printed on one of the currently available professional grade printers.The democratization of desktop 3D printers, however, now allows almost anyone with a serious interest in the technology to print models in their own office/workshop. These can be used for educational purposes and for prototyping, and represent an excellent entrée into the technology. Recently, I started printing 3D models of some of my own patient’s scans using a consumer grade desktop printer. The patient’s CTs were acquired on our Toshiba Aquillion 64 Slice CT scanner using our standard acquisition protocols. The DICOM volume data was then burned to a CD for processing. For my initial test prints, I used the Materialise Mimics and 3-matic software under their 30-day free trial period. The images from the appropriate volume were imported into the Mimics software. Thresholding is then performed to isolate the tissues in question, based on its Hounsfield units, a measurement of X-ray density. The particular anatomy of interest is then selected using “region growing” tools and a 3D model is generated. The model is then “wrapped”, to account for the individual CT slices, and to smooth any gaps in the 3D mesh. Choosing the degree of wrapping is where experience comes into play. Too little wrapping can cause gaps to be present on your final models. Too much, and detail can be lost. The 3D model is then exported into the 3-matic program for “local smoothing” of the model. The digital model is then hollowed, depending on the structure and its use. You then export it as a binary STL file. In all of the steps above, clinical knowledge of the anatomy is extremely helpful in creating the most accurate models possible. Understanding how the models will be used informs your decisions in their creation. The STL file is then imported into a slicing software to create the G-code files that instruct the printer how to actually create the physical model. I used the open source Cura software for the generation of the G-code for the printer. An image of the 3D model is seen superimposed in a representation of the build plate of the particular printer, in my case the Ultimaker 2 (Ultimaker B.V.). The model can be rotated to optimize the printing process. The highest resolution of current 3D prints from fused filament printers is in the z-direction: from the bottom on the build plate to the top of the object. The degree of overhang must also be taken into account. Since the filament cannot be deposited in thin air, the slicing software creates a scaffold to support the overhanging material. Keep in mind; this support structure must be physically removed in post-processing. The slicing software also creates a thin base layer of the material (called a brim or raft) that is deposited around the object to facilitate print adherence to the print platform. The generated G-code is saved to an SD card, which is then placed into the printer. In some cases, the files can be transferred wirelessly. I used 2.85 mm PLA filament to create the printed models. PLA is polylactic acid, a biodegradable material derived from cornstarch. PLA based material has been used in orthopedics for sutures, controlled release systems, scaffolding for cartilage regeneration, and fixation screws. The print time takes several hours, depending on the size and complexity of the model, as well as the amount of support structure used. Through trial and error, I found that careful positioning of the 3D model on the virtual build plate can potentially shorten the length of printing time. A full-scale hollow abdominal aortic aneurysm model took about 9 hours to print, while a full-scale scapula took 13 hours. A life-size pediatric skull will take approximately 23 hours to print! The use of 3D printing in medicine presents enormous potential. Exponential development of many new applications will occur if researchers, students and clinicians have access to small-scale 3D printers for prototyping new devices and procedures. The future is only limited by the imagination. A method of reimbursement wouldn’t hurt either. I would like to thank Frank Rybicki, MD, Professor and Chair of Radiology, University of Ottawa, and his team from the Applied Imaging Science Laboratory at Brigham and Women's Hospital for their great 3D Printing Hands-on courses at RSNA 2014. Copyright ©2015 Eric M. Baumel, MD
  9. Greetings! Does anyone have any suggestions for a first 3D printer. I was going to get the Makerbot, but the reviews of the new 5th generation model give me doubts. I am a radiologist with interst in making 3D models for educational purposes. Thanks for you help! Eric
  • Create New...