Hello. My name is Petar Valchanov, I’m an Anatomy and Cell biology assistant professor at the Medical University of Varna, Bulgaria. My job is to teach medical students in the trunk and limbs dissection courses, and the neuroanatomy and the histology courses of the Anatomy department’s program, to finish my PhD project (bioprinting of functional bone tissue with nanocomposite hydrogel and mesenchymal stem cells) and to provide clinical anatomical support to my colleagues from the surgical departments. I’m also responsible for “everything 3D” at my department.
When I graduated in 2004, “going Necro” wasn’t my life plan, so I started an honest job as an emergency physician at the prehospital healthcare system of Varna, Bulgaria. I had 12 years of experience as an Emergency internal physician and 3 years of experience as a psychiatry resident. One day I felt that I was totally burned out, I had seen everything that medicine could throw at me, I had had enough and I needed a quiet job, far away from the horrors of the ER.
I was really excited by the TED speech of Prof. Anthony Atala, who 3D printed the first functional bladder with biopolymers and stem cells. So I decided, that is it! I won’t treat people, I will 3D print them instead! Or, at least, parts of them… So, I left the ER, I applied for an assistant professor position at my alma mater, with the last of my savings I bought a second-hand laptop, and 4 years later I’m about to finish this little project of mine.
The necessity led me to the 3D modelling—you cannot 3D print what you cannot model by yourself. This is literally the key to make all your dreams into reality.
The website for medical 3d modelling embodi3d provided me with the necessary tools—the video tutorials of Dr. Michael Itagaki with open-source software for medical segmentation and surface sculpting. Still, I believe that those are some of the best tutorials in the field, with a training data set attached to the tutorial. I was able to create my first 3d models, then I started to perfect my technique. I had some comprehensive training in several medical specialities and 4 years later, I’m able to create a 3D printed replica of a human bone or blood vessel from a CT or MRI dataset with an average deviation of 300 microns—I can’t go below that, because I can’t break the laws of physics (actually, I CAN go better, but the radiation dose for patients would be too high and potentially lethal…). This kind of accuracy can really change the outcome of a surgical operation, so I started to make orthopedic and cardio-vascular 3D models for a few clinics and to 3D print them from plastic and rubber for presurgical planning and training. Every 3D model of mine is an actual patient, with an actual pathology and actual fate.
I’m also making 3D models for computational fluid dynamics (CFD) and real benchmark tests as a freelancer; the nasal cavities are my specialty. I also really like to segment and 3D model CT scans of animal skulls and to turn them into toys for the kids.
3D and Me
I came from a long and glorious line of Bulgarian camera-men and film-makers and in my family, everyone has “a good eye”. Actually, I’m the “medical black sheep” of the family. I never imagined I’d have to deal with something, related to my big brother’s job (he’s Kiril Valchanov, a Hollywood-level steadicam operator, the best in Bulgaria). I had some experience with coding—I started with Basic, when I was 10. Also, I modelled paper and plastic models a lot and played a lot with my 8-bit computer, when I was a kid (we’re talking about the early 90-s). I had the chance to play a lot with Robko 01 (ancient, Bulgarian-made, SCARA robot) and this is when I met the robotics for the first time. I didn’t realise that one day I would have to deal again with robots and my career would depend on it, though.
I had some experience with PhD and CSS, but nothing too impressive. But when I started to work with a 3D program for a first time (Autodesk Meshmixer, Autodesk Fusion 360), it was quite natural for me and now I consider myself quite capable in surface sculpting and capable enough in object design for my goals in medical 3D modelling and 3D printing.
I’m really impressed by Blender—it’s like the equivalent of a Swiss army knife in the CAD software and you can do virtually EVERYTHING with it. When I was trying to deal with navigation, though, I decided that I would spare myself the misery and it’s fine for me to be an “Autodesk person”.
In the medical 3D modelling world, software is quite expensive. Most of the programs cost tens of thousands of dollars and it’s impossible for an average Joe like me to work with this kind of software. So I’m working with open-source software for medical segmentation, 3D Slicer, ImageJ and InVesalius are amazing examples of it. Basically, with such software, you can convert everything you want from a medical dataset into a 3D model and then 3D print it, but it takes more effort and time. With a steady workflow, this is not a problem at all.
Science and Me
When I was a medical student, my dream was to become a psychiatrist like my heroes—Carl Jaspers and Sigmund Freud. I was always fascinated by the troubled minds, so psychiatry was the natural direction for me to go. So, when I graduated, I started to work as an ER internal physician on night shifts and a psychiatry resident during the day. It was quite hard, but I managed somehow to handle myself for three years. And then there was an incident—one of the psychotic patients attacked me and hurt me really badly. After that, it was impossible for me to continue with the residency, so I left and focused on the ER instead. After several years, I had a realisation—I’m 35 years old, I’m treating the worst possible patients, my career is going nowhere and I’m totally screwed.
My transition to science was a leap of faith for me. I tasked myself with three (not so) simple tasks: to become a well-known scientist, a capable medical 3D printing specialist, and a respected academic teacher. Those were the means to start my path into the field of biomedicine.
I bought myself a 3D printer (Prusa Mk3). This amazing, beautiful robot has served me loyally ever since. I started to 3D print and sell everything I modelled (skulls, spines, thoraces, hearts, clitores, penises, dinosaurs, some D&D miniatures, drone parts, modelling parts, etc.) and soon I had enough money for a second 3D printer (Creality CR-10S) and a cheap handheld scanner (xyz). Those printers are really cheap when you order them as a kit and build them by yourself.
I had several freelancing jobs for a very adequate pay rate and soon I was able to buy myself my third baby—Seemecnc Rostock Max—a powerful Delta with 32 bit controller and dual extruder. Now I have enough printing power to make into reality everything I need—pla, abs, petg, tpu, nylon, moldlay, single extruder, double extruder, multi print, water-soluble support. I started to print thick layers, thin layers, with or without support, with custom support… then I discovered the composite materials. It’s been quite a journey. 🙂
My models are state of the art—“conference models”, as the surgeons call them… Colorful, shiny and beautiful representations of something horrible, which happened to somebody. Not a name or a face, just a number. A job. Recently, I started to print with Silk PLA (a composite of 80% PLA and 20% polyester) a lot, because of its minimal material deformation after the polymerisation. This material has less than 0.2% of deformation after cooling and I highly recommend it for accurate prints. Also, it’s dirt cheap and looks quite shiny, the finishing is like a really fine silk…
The anatomical 3D models are very useful, when they are used for preoperative planning. The surgeons are relying mostly on their hands, the tactile sense and the agility of their palms and fingers. The 3D printed anatomical model gives them the opportunity to explore the spatial properties of the model and to plan better the operation window—the position of the incision and the angle of the tools. This is especially important for major surgeries, in multiple cavities of the body and combined with significant anatomical variations. You can’t just cut the patient wherever you like—there are specific windows, with specific positions of the entire field. It’s like a really bloody 3D scene, with limited visual contact, in which you have to orient yourself with your fingers and then to proceed with a dangerous procedure on a structure, which is wired and ready to blow. Also, the position of the internal structures can vary a lot (the position of some blood vessels can have more than 30 different variations) and this can lead to major intrasurgical complications (e.g., bleeding from a blood vessel, which is not supposed to be there). The 3D model demonstrates all those variations, which helps the surgeon to choose the optimal approach for the specific patient.
In vascular surgery, the models can be 3D printed with elastic materials and can be used as a training model—operation before the real operation on a phantom. For example, this 3D model is made from the Angio CT dataset of a male with an aneurysm of the abdominal aorta. The same patient also has aneurysms of both common iliac arteries, and on the left side there are 3 kidney arteries for the left kidney (usually there is only one kidney artery). Also, the left kidney is in contact with the aneurysmal sac. Without proper planning, the operation has a high risk of complications. This is why I printed the model with polyurethane rubber, real size, and sent it to the surgeons. They prepared the endovascular prosthesis and trained on the plastic of the kidney arteries, using the 3D printed model. Especially important is the anchoring of the proximal part of the endovascular prosthesis on the wall of the aorta and the rubber model is really good to train for the actual procedure. The surgery was successful and the patient is still alive and kicking.
Another example of the clinical value of 3D models is this case—a male with aneurysm of the abdominal aorta and horseshoe kidney—in some rare cases, the kidneys are not separated during ontogenesis and they remain connected in the postnatal life. In this specific case, the aponeurosis of the mesenteric arteries was entwined together and the surgery was a high-risk one. I made a model before and after the operation and I 3D printed them both with red and white PLA. The model was used for presurgical planning of the procedure and for demonstrational purposes for medical students and at scientific symposia.
In orthopedic surgery, the repositioning of multi fragmental fractures is performed by metallic osteosynthesis—needles, pins, screws and wire are used to bind the fragments together into a similar-to-normal anatomical position. This usually takes a few hours. At the same time, the patient is on the operation table, under general anesthesia, and there is an X-ray machine, which irradiates the patient and the operation team from time to time. There is an invisible clock in the operation theatre—when the patient is intubated and the operation is on, it has to finish somehow. The longer it takes, the greater the risk of complications for the patient, the radiation dose for everybody grows and the final outcome of the operation is worse. When there is a 3D printed model, the orthopedic surgeon can prepare the elements of the metallic osteosynthesis—the actual pins and needles are bound in their best positions, the length of the bolts and screws is prepared, the entire logic of the metallic construction is planned. Then the tools are sterilized and the actual operation can start. If the surgeons don’t have the opportunity to plan the operation, they have to make it from the first attempt, on the actual patient, in a high-risk procedure. The preoperative 3D model can decrease the actual intraoperative time with more than 2 hours, which significantly changes the outcome for everyone involved. So I started to print ankle models for the university orthopedic clinic in different colors and they are very good teaching tools for the medical students.
In otorhinolaryngology, the 3D models can help a lot with surgical planning. A detailed model of the airways can be made and CFD analysis can be done, similar to how it is done for wind tunnels in aerospace engineering. Such analysis can demonstrate in which position of the airway the air turbulence is highest, hence the optimal operative procedure with maximum efficiency can be chosen by the team.
I have several tasks right now:
1. To finish my PhD. For this purpose, I have to mix two polysaccharides and a nano silicate clay into a hydrogel, to add growth factors and to 3D print a scaffold with a mesh of channels into the model. For the printing I’ll use Hyrel 3D Hydra 16A. A cell cultivation media will circulate through the channels, which will be used for the inoculation of bone marrow derived mesenchymal progenitor cells. Then I’ll incubate the thing for 28 days (its name is B.L.O.B. Biosynthetic stereoLithographic Osteogenic Bioimplant. My science supervisor told me it’s a bad idea to name it, but I can’t help it…). I’m still thinking about the geometry of the channels—Rectangular seems too dull, I prefer Gyroid or Voronoi. Anyway, I have the materials and the printer, I hope to finish the experimental work in the next 6 months.
2. To make a multi-purpose surgical simulator. I’m using an Angio CT scan of a caucasian female with dilatation of the aortic arch and bifurcated rib on the left side, 1mm slide thickness, bone and soft tissues windows. It was supposed to be “a pristine thorax”, but the patients always have some anatomical variation to throw at you… I made the initial model and now I have to design three surgical accesses—for the mitral and aortic valves, also an access to arteria thoracica interna for cardio-coronary bypass. The surgeons want to simulate the access with retractors between the ribs, so I’ll have to make six pairs of hinges, two for each surgical access. In the center of the thorax will stand a model of a heart, with replaceable silicone mitral and aortic arches and rubber arteria thoracica interna. When I’m done with the design, I’ll print the thorax with PETG (Chemical resistant, can be wiped with alcohol), the heart with PLA, and sockets for the replaceable parts. I still have a lot of work on this thing, but the model is done.
3. To organize a 3D printing facility. Recently my department won funding for a bioprinter, so we bought the Hyrel 3D Hydra 16 A—a huge professional machine, which prints everything except metal (we managed to print a rectangle with hair gel out of the box). Now my colleagues and I have to organize a working facility for this thing. We’re planning to use the printer for in-house experimental hydrogel, PEEK and Polyamide implants and organ-on-chip devices.
4. To organize a “3D print yourself” course for the general public. I have made some progress with my handheld scanner—I scanned some of my colleagues, myself, several students and a cadaver, and I managed to mix a sculpture of myself as a Roman. Now I’m participating in the organisation of a course: how to acquire a working 3D scanner and a 3D printer for less than 500$ and then to use them with open-source software to scan and 3D print actual people with PLA plastic. Sounds fun to me, I hope we’ll have enough participants. 🙂
5. To print more protective face shields for the local hospitals. We’re in the Covid-19 outbreak, there is an emergency situation, and there is not enough protective equipment for the guys from the ER. So, I organized a little face shield manufacturing. Me and a maker buddy of mine are printing one of the designs from the National Institutes of Health website. We’re printing it in PETG, on 0.2mm layers, with a 0,2mm thick PET sheet for the shield. I highly recommend PETG and PET because of their chemical resistance and their ability to be wiped multiple times with alcohol without major structural damage. If you have a 3D printer, follow the initiative, print some face shields, donate them to a hospital—this really helps.
I usually start with a phone call from a surgeon about a new patient in the ER with a problematic condition, which requires my services. The surgeon explains to me what is the most troublesome part of the patient’s condition and gives me the region of interest (ROI), which I have to illustrate. In the next few hours, the surgeon sends me the data set and I can start with my work. If I have the option of a contact with the radiologist prior to the study, I can request a special computational kernel for the raw data set, but usually don’t have that option.
When I get the data set, I take an hour to observe the data set in a dicom viewer (I’m really happy with Radiant). I’m planning the exact structures, which I have to segment next.
Then I begin with the Segment editor of 3D Slicer. First, I add several control voxels at standard anatomical positions, usually ostiums of blood vessels, bony landmarks or remaining metallic implants (You would be surprised how much metal you can find in an average adult person…). This gives me the parameters of the 3D scene and I use those voxels in the following quality control.
When I’m ready with the control voxels, I start with the segmentation of the structure of interest. Every structure in the body has a different value in Hounsfield units. The water is 0, the air -1000, etc. If I have to segment a bone, I’m using the respective range of HU (250—+1000). In this way I have a segmentation of the voxels in the body of the patient with a specific HU value (label map). Finally, I export the label map as a .stl file, which manages to save the spatial coordinates of the voxels and to combine them in a triangular mesh of vertices and polygons.
This is where the sculpting begins. I use Autodesk Meshmixer, because it’s simple, free, has a built-in slicer for 3D printing and an amazing tree-like custom 3D printing support. I don’t use textures, so ZBrush would be an overkill for me. For more complicated operations (hinges, magnet sockets, holes, joints etc.) I use Autodesk Fusion 360. The complicated operations occur during the postprocessing of the model. I remove all the artefacts, close the gaps and smooth the model, until it’s a clean shell, representing the external shape of the object of interest. Then I import it again in Slicer 3D, converting it into a label map and I begin the detailed work on the segmentations. In this way, I can segment even the most complicated and small structures, if their size is at least one subsequent voxel in three subsequent slides. This means thinner slides and better details. 1 mm slides are good enough for most of the bones and the blood vessels. With the control voxels, I can check if the model deviated too much during the initial modelling.
When I’m done with the second segmentation, I export the model again for final touches on the retopology and postprocessing of the final model. I upload the model to Sketchfab and send the link to the surgeon, so the team can start with the planning. A usual aortic aneurysm takes 3-4 hours; an ankle takes 5-6 hours, if the fracture is not compressive.
In the next 3 days I 3D print the model and when I’m done, I send it via a courier. The preparation for a complicated surgery takes at least a week, so my model arrives just in time.
As a quality control, I measure 6 distances between the control voxels on the dicom viewer and on the 3D model; then I calculate the average deviation. If it’s under 0.5mm, I’m good (it usually is). On a CT scan with thin slides (0.7mm or better), I can achieve an average deviation of 0.3mm.
Sketchfab and Me
When I made my first 3D models, I had a dire need to present them properly in department meetings, to my students and to be able to send them easily via email or social networks. First, I tried a portable 3D viewer, but it was too inconvenient and unpractical. Then, I installed a generator for 3D PDFs, but the engine was slow and clumsy, especially with my high-polygon models. And THEN I saw Sketchfab for the first time. It was magical! Large files to upload, cool scene effects, material properties, annotations, model analysis. It was the perfect solution for my needs and I have been using it ever since.
Sketchfab is a really nice help in my job as an anatomy teacher. There are amazing anatomical models, with texture and annotations, made by some of the best 3D artists in the world. I’m using them constantly in my seminars. During the coronavirus outbreak, I have to participate in distance learning with video-chat and those models are life-saver. My students like them a lot and they use them regularly. Last year I was able to create an elective course “Virtual anatomy and 3D modelling” for medical students. So far, my students are a bit clumsy, but I’m sure they’ll become great 3D artists one day. 🙂
Sketchfab is a great way to present medical 3D models at pre-surgical team meetings, scientific congresses, or in presentations. For me, it’s a way to access my portfolio from everywhere, when I need it. Also, it looks quite catchy on a transparent multi touch screen, which is embedded in a rack with actual human hearts. 🙂
My Advice for the Newbies
- 3D print thick layers, 3D print thin layers, with or without support or with custom support…then discover composite materials. Most of all—3D print and you’ll see how your dreams are materializing before your very eyes, layer by layer.
- Build your portfolio constantly. If you can’t add your NDA projects to your portfolio, add them to your CV. Your portfolio and your CV—this is who you are in the eyes of everyone else.
- Revisit your old models from time to time. It’s a good way to see your previous mistakes and to test your new techniques.
- Acquire a PhD. This is the entrance level to the real cutting-edge technologies. Then you can continue in academic or corporate fields, the choice is yours. A whole new world will be accessible to you.
- Don’t be a Yes Man.
- Teach somebody about something. By teaching others, you’re teaching yourself.
- Do stuff with your hands constantly. If you’re good with the scalpel, you’ll be good with the cursor too.
The main goal of science is to make something uncertain into something certain. 3D technologies are a powerful tool for image science and morphological analysis of macro and microscopic structures. The combination of additive manufacturing and CNC milling, makes possible the building of three-dimensional structures with complex geometry from virtually every possible material, which will push forward every aspect of our lives. I can say for certain that it’s beyond our wildest dreams.
Simple biosynthetic organs are made right now and there are major breakthroughs in the bioprinting of complex vascular structures, which is the requirement for complex artificial internal organs. A functional myocardial tissue was created several months ago and the new nanomaterials make the impossible possible, regarding wound healing and the regeneration of complex anatomical structures. Soon we will be able to rebuild ourselves, harder, better, faster, stronger… or at least those of us who can afford it. We’re walking toward a world without disease, impairment and pain, but I don’t believe the current economic system will make those amazing technologies accessible to everyone.
For my institution, the future looks quite bright in the field of regenerative medicine. We have all the necessary means to start several research projects with some of the most promising materials for medical implants and devices (experimental hydrogel, PEEK and Polyamide implants and organ-on-chip devices). It will be a bumpy ride, which will take years, though.
I’m planning to increase my teaching hours in 3D and to train some students for medical data segmentation and surface modelling. I hope I’ll inspire their interest in medical 3D modelling and printing, clinical anatomy and radiology. Better doctors, better world, right…
For now, my goal is to keep doing my tasks and eventually to become an unforgettable part of the future, for good or for bad.