top of page

Meticuly insights: Combining AI and cloud-based surgical planning for a smooth user experience

Machine learning and artificial intelligence (AI) frequently face skepticism regarding their practical applications, largely due to the challenges in translating academic research into effective real-world solutions (Bloom et al., 2017). One area where these technologies have shown promise is in the manufacturing of patient-specific implants, particularly in overcoming the complexities associated with image segmentation. The segmentation process, which involves delineating anatomical structures from 3D DICOM scans, is often a laborious and time-consuming step (van Eijnatten et al., 2018). To address this, Meticuly has investigated the potential of deep learning algorithms to significantly streamline and automate the entire of the design and manufacturing processes; with this exploration aiming to demonstrate how advanced AI techniques can enhance efficiency and accuracy in the production of custom implants, ultimately bridging the gap between theoretical research and practical application. 


Traditionally, the process of implant design for various craniectomies involves imaging, data processing, design, and validation before production. Among these phases, data processing, anatomical reconstruction and implant design are particularly cumbersome, as the editing must be performed manually, with each pixel being labelled as either bone, cartilage, or other tissues. This manual identification is prone to inconsistencies, as distinguishing between bones, tissues, and miscellaneous particulates from still imagery can be challenging. Moreover, reconstructing the bone requires skills and time to perform and appropriate quality control to ensure that the implant design accurately fits to the patient anatomy. AI has the potential to address these issues by automating the design process. Through training on large datasets, they learn to differentiate complex structures like bone, cartilage and the soft tissue with increasing precision of generating bone loss and the cranioplasty design. By leveraging deep learning techniques, AI streamlines the data processing phase, reduces human error, and enhances the overall precision of implant design, ultimately improving the effectiveness and efficiency of creating patient-specific implants for cranioplasty


​​Building upon these advancements, Meticuly has developed its own tailored CraNeXt algorithm, derived from 3D U-Net architecture, optimized for cranial reconstruction procedures. This specialized model is trained on a comprehensive dataset, allowing it to achieve greater precision in reconstructing bone loss. An additional benefit of Meticuly’s solution is its ability to accurately determine skull thickness by comparing the patient’s scan with its training data. This precision allows for more informed decisions on where to place screws that secure the mesh to the skull, particularly around the edges of craniectomy openings, ensuring a better fit and improved stability (Kesornsri, T. et al., 2024, Wanavit, T. et al., 2022).


This algorithm is built into our case management application, thereby facilitating the time required to create the preliminary design of the cranioplasty for design planning. The principles of the AI algorithm presented in this article, can be applied to any product, making Meticuly’s patient-specific implants much more accessible. 


Meticuly Application Combining AI and cloud-based surgical planning for a smooth user experience

References

Bloom, N. et al. (2017). Ideas aren’t running out, but they are getting more expensive to find. https://voxeu.org/article/ideas-aren-trunning-out-theyare-getting-more-expensive-find.

Kesornsri, T. et al. (2024). CraNeXt: Automatic Reconstruction of Skull Implants With Skull Categorization Technique. IEEE Access, 12:84907–84922, doi: 10.1109/ACCESS.2024.3415173.

van Eijnatten, M. et al. (2018). CT image segmentation methods for bone used in medical additive manufacturing. Medical Engineering & Physics, 51:6–16.

Wanavit, T. et al. (2022). Technology transfer of Convolutional Neural Networks: An example. Proceedings of the 14th International Joint Conference on Computational Intelligence, pp. 375–380. doi:10.5220/0011540200003332. 


Comments


bottom of page