UK Researchers Inspired by Astrophysics to Improve Imaging for 3D Printed Models

Like cancer, heart disease and the many conditions surrounding it (along with other related systems in the human body) can often cause fatal complications, leading researchers around the world to continually look for better methods in diagnosing, treatment, and surgical procedures. UK researchers, I. Brewis and J.A. McLaughlin, at Northumbria University have recently explored 3D imaging in reference to cardiovascular health care, publishing their findings in ‘Improved Visualisation of Patient-Specific Heart Structure Using Three-Dimensional Printing Coupled with Image-Processing Techniques Inspired by Astrophysical Methods.’

Brewis and McLaughlin are working in the realm of astrophysics in creating new image-processing techniques for viewing the human heart, transferring the data to an .stl file and then 3D printing a medical model. These new techniques allow for better modeling from scans, especially improving on the clarity of smaller features.

Traditionally, CT and MRI scans have supplied data for 3D printed models. As the authors of the study point out, these are ‘relatively accurate’ but still produce errors. Not only that, image processing can be a high-maintenance venture, consuming both time and money. Currently there are a range of techniques in use, including:

• Vignetting
• Boxcar smoothing function
• Dilation
• Edge detection
• Pixel plate scale development for charge-coupled devices on spacecraft

Brewis and McLaughlin examined whether using techniques generally used in astrophysics could reduce the margin of error, along with faster turnaround time in rendering. The basic steps in arriving at a 3D model must include acquiring data, segmenting, converting files, fixing and design, and 3D printing. For the purposes of this study, the scientists used data from one anonymous patient, obtaining 856 CT images. In segmentation they were able to identify the ‘region of interest’ to be 3D printed: the heart tissue.

Segmented heart data following first thresholding of right heart region (left) and segmented heart data following second thresholding of right heart region (right). The right ventricle (bottom left) shows improved clarity of internal chamber segmentation following second thresholding

The team used Slicer to view the chest cavity from multiple angles and cut all extraneous data for a better view.

“Initial thresholding highlighted all heart tissue, with the inclusion of deoxygenated blood in both the right atrium and right ventricle,” stated the researchers in their paper. “Areas of deoxygenated blood were found to be removed by applying a second threshold in the range [-600, 70] to the chambers of the right heart.”

A second pass of thresholding resulted in converting data from 2D segments into a 3D rendering.

“This method, whilst modelling the majority of the heart’s internal structure, omitted small-scale features such as valves. In order to extract these small-scale features from the DICOM data available, an alternative approach was introduced,” stated the researchers, upon using vignetting to eliminated added interference in the background.

Netfabb image processing stages from left to right initial and divided models

Further highlighting and enhancement allowed the research team to see areas of aortic valve cusp occlusion and areas of bad pixels more clearly.

“For both the full heart and the aortal valve, the total time taken for image segmentation was on the order of tens of minutes, where some parts of the segmentation process were, naturally, more time-consuming than others,” stated the researchers. “For the method presented here, the main time-consuming step was whilst utilizing Slicer 4.8.1’s in-built eraser and draw tools.”

Once the researchers had the files for both the heart and aortic valve models ready, they used Netfabb to prep for production with an SLA 3D printer.

“The SLA printer produced more sturdy 3D models yet could not produce models which did not require additional support in order to maintain the desired structure of the internal heart chambers (i.e. free standing models) during printing,” stated the researchers. “For a simple and relatively flat structure however, such as the aortic valve, the free standing issue was not a concern and the stronger resin-based model proved to be ideal for producing thin and delicate structures such as the tricuspid valve.”

Ultimately, the team produced a 3D printed heart model with four clearly defined heart chambers, along with the aorta, superior vena cava and the pulmonary vein and artery. The patient whose data the researchers were using as an example had an aortic aneurysm that is accurately depicted in the model, seen as an area of depression in an aortic side wall. Calcification can be clearly observed too, and this type of accuracy means surgeons can operate more precisely and efficiently, saving on time in the operating room.

“The use of 3D renderings of patient data improves on traditional imaging techniques where surgeons are required to visualise a three-dimensional picture of heart defects based on a series of 2D scans by reproducing exact, real three-dimensional cardiovascular anatomy,” state the researchers. “The use of 3D modelling can also improve the physician’s understanding of individual patient anatomy such as in the case of valve replacement16 or in procedural planning for the treatment of congenital heart disease.”

What do you think of this news? Let us know your thoughts! Join the discussion of this and other 3D printing topics at 3DPrintBoard.com.

3D printed heart models. (left) complete heart model, (right) aortic valve
model). UK one pound coin for scale.