91³Ô¹ÏÍø

Segmentation in Simpleware Software

There are several different options for segmentation in Simpleware software when working with knee CT data for a project of this type. Manual image segmentation involves carrying out different operations on the image data, including cropping, thresholding, and apply a mask floodfill to remove unwanted parts and obtain the required anatomy, in this case the tibia. The ¡®Split regions¡¯ tool also uses algorithm markers on different regions to split image masks and isolate the tibia. In addition, tools such as ¡®Close¡¯ and ¡®Cavity fill¡¯ can solidify the mask further, prior to smoothing and creation of a high-quality triangulated surface for export to nTopology.

However, these manual steps can also be automated using a range of basic and more advanced scripting options. The first of these approaches involves recording a macro during manual segmentation to show and run all functions/operations ¨C scripts can be edited in Python and C#, including the removal of any workflow steps not needed or unavailable in scripting, such as split regions/painting on markers. By replaying the script, the same output can be generated. Taken further, a script can be used to create a plug-in to be added to the Simpleware ribbon, which can be set up to allow for the macro to be run in different stages, depending on the manual operation that still needs completing (such as split regions).

AI-Based Machine Learning Automation

To extend automation, the AI-based Machine Learning module Simpleware AS Ortho/CMF can be used to carry out the segmentation in approximately two minutes, with scripting then enabling any additional steps to be carried out. The same module can be used to add landmarks as point measurements to the segmented tibia for designing a surgical guide. 

For this project, around 50 patient-specific tibia models needed processing, with the Simpleware AI solution meaning that a large workflow could be rapidly completed without repetitive manual segmentation. If needing to have all the tibias in approximately the same location, the landmarks can be used as reference points for registration tools to bring all the tibia surfaces together.

Design Process in nTopology

The data exported from Simpleware software was then used as input to the reusable design process created in software. nTopology enabled the team to automatically generate the conformal geometry of the mating surface of the guide from the patient-specific model and parametrize the location of the cutting holes and slits based on the landmarks provided by Simpleware software. 

Furthermore, the design workflow included manufacturability considerations, the implementation of structural support columns, a flush base structure, and venting features. The output of the design process was a mesh file that was ready for additive manufacturing.

Designing this fully featured end-to-end workflow for the tibia surgical guide means that clinicians or technicians can easily implement changes, such as adjusting the guide¡¯s design and customizing cutting angles for different patient scans with a specific set of landmarks. Since the expertise-intensive manufacturability considerations are already encoded into the design process, the resulting workflow was easy to use even by users without engineering or design experience.

Automating Design Generation with Scripting

Using this approach, there is a clear path to scaling up and automating design generation of patient-specific devices. Once the design process is created and validated for a single cutting guide, you can then automatically batch process a whole folder of patient scans using simple scripts and nTopCL - nTopology's command line interface.

In this proof-of-concept, the team created a Python script to process the 50 tibias and automatically generate a unique patient-specific guide for each of them. Spending a few minutes to write a few lines of code saved hours of manual design work. For more complex projects, the same method could be applied to build out complex logic into the workflows that can combine data from multiple sources - like simulation results.

Conclusion

This joint workflow developed using Simpleware software and nTopology reduces time and manual effort when working on projects with large amounts of patient-specific anatomies. While this particular project focused on a tibia cutting guide, different anatomies can be selected either with the Simpleware anatomy-specific Auto Segmenter modules, or a fully customized workflow. As well as making it possible to scale up surgical guide design with patient-specific data, this approach uses automation to cut down on the technical expertise needed when working with advanced software.

Learn More

• Watch a recording of the Synopsys-nToplogy webinar , demonstrating the segmentation and design workflow from this case study.
• Download a copy of the nToplogy Guide on .