Abstract
We appreciate the thoughtful comments from Amnuay Kleebayoon and Viroj Wiwanitkit to our contribution, “A Low-Cost Workflow to Generate Virtual and Physical Three-Dimensional Models of Cardiac Structures.” In their comment the authors raise a valid concern regarding the precision, cost-efficiency, and usability of the proposed workflow, specifically whether it is sufficiently accurate, affordable, and easy to use for medical professionals with limited expertise in medical imaging or 3D modeling.
While modern computed tomography scans can achieve a z-resolution of 0.5 mm or better, magnetic resonance imaging scans typically have a slice thickness of 1 to 2 mm, making them less suitable for 3D segmentation. The printer used in this workflow costs approximately €200 and is marketed with an X/Y precision of 0.0125 mm and a Z precision of 0.002 mm. This indicates that printing resolution is secondary to the resolution of the primary imaging, emphasizing that even low-cost hardware can be used with sufficient precision.
Secondly, it was noted that many physicians may lack the technical knowledge required to implement the workflow. While it is true that these skills are not typically taught during medical specialization, all the steps described in our workflow were successfully carried out by physicians and pregraduate medical students (under the supervision of a cardiac surgeon) at our institution, without prior expertise. Once the workflow is established, the necessary skills for segmentation and model editing can be acquired with ease. The software described in the workflow includes semiautomatic segmentation tools, which aid in the process. As the authors correctly point out, commercially available software also offers fully automated, AI-supported segmentation tools. While these can accelerate the process, they come with significant financial costs.
Lastly, the authors discuss both the direct and indirect costs associated with this low-cost workflow, including the operational expenses of 3D printing, post-processing equipment, and the training time required for medical professionals. After several hundred hours of printing with the described low-cost printer, we only needed to replace the nozzle once, which incurred a cost of approximately €5. Throughout the process, we used affordable printing material at a cost of €20 per kilogram, with each model consuming between 20 and 50 g. Since no post-processing was required for our approach, there were no additional costs. We hope that as 3D technologies become increasingly accessible, their integration into clinical practice will be facilitated through approaches like ours. We believe that incorporating 3D modeling into medical education—just as surgical skills and the use of medical imaging are currently taught—will contribute to further optimizing patient safety.
