Latest Team meeting 2 - old office

Automation and Internationalization: Our founder looks ahead to the next twelve months at Axial3D


From automation to internationalization, our founder, Daniel Crawford, gives us his thoughts on where Axial3D is headed, and what the next 12 months look like in the land of medical 3D printing.

Over the last year... No, let me start over. Since the beginning at Axial3D, we’ve developed and grown at a rapid pace. Constantly evolving, learning and growing to achieve one thing. To make it easier for clinicians, all around the world, to gain access to 3D printing.

As I look back on 2018, it feels like we have laid the foundations to explode into the market in the next 12 months. Last year, the team grew by 75%, we established 3 longstanding relationships with Universities, and we grew into the US markets with the first of many hospitals signing on to use our services to bring 3D printing closer to clinicians.

When I look forward to the next 12 months at axial3D, there are two main themes for 2019 and beyond. Automation and internationalization.


From the beginning, we have seen a marked increase each month of surgeons requesting 3D prints of their patient's pathologies. Given that the models are physical, tactile representations of their own patients, their benefits are instantly recognizable when compared to traditional 2D imaging and planning.

From the company’s perspective, that the 3D printed model benefits were instantly known to surgeons was fantastic. However, to meet the growing demand we needed to get creative. It was obvious that the ‘traditional’ methods of medical 3D printing would not scale in local markets. The vast majority of surgeons were too stretched to do their own processing of data (segmentation) so outsourcing was their best option. With hard copies of data on CD’s being posted, prescriptions being collected sporadically via any number of channels (such as email, letter, dictation & text messaging), it was never going to be routine.

After analyzing clinical workflows and how we would best fit into them, we created axial3Dinsight. A simple web application that linked directly to their PACS and allowed quick and easy data and prescription transfer. This meant surgeons could now order models in minutes. This platform has helped us scale within our local public healthcare markets, within the HSE and NHS, rapidly over the last 24 months. This scaling has allowed axial3D to unearth another, and arguably the most important issue, in making the process of 3D printing routine in medicine. Segmentation.

Segmentation is the process of taking medical images (DICOMS) and delineating specific structures & anatomy within them. Depending on the complexity of the model and image type, this could take anywhere from 30 minutes to 5 hours of work per 3D print. For us to scale effectively and meet demand, we had to scale a skilled team. This is a challenge for any SME & even more difficult if you are within a hospital. When we evaluated the digital thread from 2D image to 3D printed model, we found that the process of segmentation was the main reason a lot of hospitals were not adopting or scaling 3D printing internally. It became obvious that this was the next bottleneck the company had to fix.

The static portion of the image represents a cardio-thoracic magnetic resonance image. The dynamic portion corresponds to the confidence with which a machine learning model believes there is blood in a corresponding portion of the static image. The model, which learns by example to locate blood within many different magnetic resonance images, slowly starts to identify the blood more accurately within this image, which is never allowed to be used for training. This is a representation of the evolving ability of the model to accurately identify blood within the given image.

Using Machine Learning we have begun developing a completely automated approach to segmentation. These algorithms are formed on the basis of gold standard data, effectively learning how to segment & delineate certain anatomy from differing scanning modalities. As with all algorithms, with the words ‘AI’, ‘Machine Learning’ or ‘Deep Learning’ attached, the most important aspect is having the data, and lots of it. Through collaborations with University Hospitals and gaining access to the UK Biobank we have been able to collate millions of data points to hone our automation efforts, with a view to collect the worlds largest training dataset for medical 3D printing in 2019.

Initially being used internally by our team, we have verified the process and reduced the time taken to create 3D prints. This means that we can produce more models without scaling our team to an unmanageable size.

Over the next few months, we will be releasing these automated algorithms into the wild, working initially with our already established partnerships with hospitals across the UK (Barts Health NHS Trust), EU (Universitätsspital Basel) & US (Tallahassee Memorial HealthCare). Our belief is that surgeons will think of best use cases for 3D printing in medicine, so with an easy to use application, axial3Dinsight of course, as well as eliminating the labor-intensive process of segmentation, it opens up a whole new breed of user. Meaning that 3D printing in medicine can start to be used en masse & reach its full potential.


From spending time in international markets, it is clear that the company needs to focus on the USA. From working within the RSNA (Radiology Symposium of North America) 3D printing special interest group & recent reports such as the SME 2018 stating a growth of 3200% in medical 3D printing in the US its a natural market for the company to focus. The USA’s commitment to improving care with 3D printing is backed up by the FDA, who has recently released guidance on 3D printed medical models, and perhaps most significant of all is the American Medical Association’s announcement on CPT codes being approved by the AMA. From holding discussions with some of the longest established labs in the US, we’ve come to realize that they face the same issues as the company. Huge interest and need for 3D printing internally in a hospital, but no solution to scale effectively.

3D print labs such as Mayo Clinic and the VA have been running for over a decade helping to improve patient care within their healthcare systems. However with the recent desktop revolution that has been ongoing since 2013, over 100 labs (& counting) now exist within the US alone. And, all having the same barriers to scaling. With the vast majority of work being carried out by Radiologists, Surgeons & Medical Engineers it is difficult for them to scale effectively.

Our recent partnerships with both Amazon Web Services and Google Cloud allow us to leverage this distributed infrastructure to allow the company to quickly train & deploy algorithms in labs, while still ensuring data compliance is upheld.

Utilizing our cloud infrastructure, the company is set to scale in the US using axial3Dinsight to manage the collection of data, and axial3Dassure to manage the quality of output in a lab. These platforms coupled with our evolving automated algorithms will effectively allow any hospital to start, or scale their lab extremely quickly, as and when their demand increases. Ensuring that they can produce enough to meet the ever-increasing demand, as well as reducing the time clinicians spend on labor-intensive tasks, allowing them to focus on what is important... the patients.

We are always on the lookout for new and exciting partnerships with healthcare institutions that will help us to fulfill our vision of routine 3D printing in hospitals. So if you would like to hear more about our products, or discuss collaborations, feel free to reach out via the form below!