Total hip arthroplasties (THAs) that include patient-specific implants, guides and surgical plans have been rapidly increasing in frequency over the past decade. This increase is creating more demand from patients, and increasing pressures on clinicians and medical device companies to deliver and deploy their patient-specific products quickly and safely. In this context, anatomical data from MRI and CT scans can be used to create 3D models for 3D printed orthopedic cutting guides, implants, and pre-surgical plans. The Synopsys Simpleware group and the Corin Optimized Positioning System (OPS) group have been working together for over 10 years to develop and streamline their OPS workflow, which is designed to optimize stability, longevity, biomechanics, and patient outcomes for THAs over time. The result of this collaboration is a deeply integrated and customized solution for different stages of the OPS workflow, which has been deployed to over 20,000 patients, and has since evolved to include Machine Learning-based AI technologies. How, then, did this come together, and what does it tell us about the current state of optimizing patient-specific workflows? This blog serves to elaborate on our thought process regarding this integration.
Solving Workflow Challenges
The primary challenge when optimizing patient-specific workflows is to ensure speed, efficiency, and scalability, while also maintaining clinical accuracy and regulatory compliance. When creating 3D models of a patient’s anatomy, it is typically left up to a group of highly-trained and valuable biomedical engineers to complete this task; it is therefore particularly important for a company to deploy a highly efficient and scalable process, free up as much engineering time as possible, save on costs, get to market faster, and maintain the quality of the final output.
Software solutions have been developed to make it easier to import and process 3D models from imaging data, including regulatory areas, to validate processes, reducing the reliance on physical testing. In addition, open-source and commercial software programs now offer a wide range of options for visualizing, analyzing, and segmenting patient-specific imaging datasets in order to export models to design, simulation, and 3D printing applications. Reducing reliance on manual work is crucial to making these virtual methods a commonplace option for physicians.
The growing use of machine learning in medicine has so far focused primarily on detection algorithms, such as for detecting cancer or other pathologies on images, but it is also now being developed to account for other types of software challenges. These challenges can include image segmentation (or labeling) and landmarking for applications like 3D patient specific surgical planning and guide design.
In addition, open questions remain over regulatory factors including validation and algorithm control. According to FDA, a ‘locked’ algorithm is defined as one that ‘provides the same result each time the same input is applied to it and does not change with use.’ By contrast, ‘an adaptive algorithm (e.g., a continuous learning algorithm) changes its behavior using a defined learning process.” The risk then becomes how the latter algorithm can be quality controlled.
Development In-house or via Vendor?
When deciding to deploy a software solution for patient-specific workflows, the stakeholders need to decide how much of the solution, if any, can be built in-house with existing/acquired resources, and how much needs to leverage third-party vendors with off-the-shelf and/or customized solutions. The advantage of in-house path is that the entire development process and all Intellectual Property (IP) can be controlled with all added value retained. This in-house development can be done in groups ranging from entire R&D Business Units in large companies, to a few developers in small startups.
However, this approach results in significant overhead and admin expenses, as well as technical challenges that a typical medical device company may not be fully prepared to handle. Admin challenges can include finding, hiring, and retaining the right expertise and talent, especially in a “hot” area such as machine learning and AI.
Technical challenges can include deploying, updating, and maintaining the solution while ensuring all regulatory requirements are met. Although deploying a solution from a third-party vendor may have vendor-dependent higher upfront expenses, the complete solution is typically deployed in a much shorter time frame. The result is also setup to meet any regulatory requirements and supported by a team of industry leading experts, saving time/money in the long run.
Putting Together a THA Workflow
In the case of Corin, their requirements are to process 3D image data for its OPS system, and to make this workflow fast, scalable, and accurate. The first step for this workflow involves the operating physician ordering CT-scans of the hip, knee, or ankle. The image dataset is then imported into Simpleware ScanIP software.
Over the years, image processing has incorporated manual off-the-shelf workflows, scripting plugins, and now AI-based software solutions that automate common processes such as obtaining anatomical regions of interest and landmarks, with the latter used to help with implant orientation.Once complete, a few corrections are needed to optimize the results. An automated approach is also suitable for batch processing using “console set-ups” in the software to maximize throughput. By leveraging the software’s scripting API (available in C#, Python), Corin was able to develop custom plugins to further streamline and customize their workflows.
Caption: Typical relative processing time savings for Hip, Knee and Ankle combined (segmentation and landmarking) when compared to using off-the-shelf software with no customization are the following:
- Using non-AI scripting plugins *: Reduced time by 65% per case
- Using Custom Modeler plugin: Reduced time by 94% per case – ~2/3 of time is computer time (no human interaction) and remaining is tidy up time
*Plugins developed using C# scripting API
Using models for implant templating
Corin creates 3D surgical templating conducted on its OPS Insight online planning platform. Once the surgeon has reviewed and validated the planning, patient-specific implant guides are 3D printed to be used as an intraoperative delivery system for the clinician to achieve their target implant positioning.
Achieving surgical results
The use of Corin’s OPS solution has gradually expanded worldwide, and it is currently used for approximately 6,000 cases per year, with the milestone of 20,000 total cases passed earlier this year. In addition, OPS is now available in 13 countries and used by almost 280 surgeons around the globe. The OPS technology has also been the subject of 26 peer-reviewed publications, supporting the clinical relevancy and value of this technology. In addition, projects are being carried out by other organizations, including 3D printed patient-specific models for complex hip arthroplasty, and general adoption of AI for pre-operative planning, as developers, clinicians, and medical device companies test out what is possible with combining different technologies.
Future challenges and opportunities
As the use of patient-specific workflows for device design and surgical planning grows into different anatomies and medical device applications, having access to flexible and customizable end-to-end solutions will be crucial to future innovation and growth for medical device companies.
For the case presented in this blog, there are many more challenges to be met when it comes to optimizing patient-specific data and virtual surgical planning. The goal is to have the fastest turnaround time without compromising on quality and accuracy, a process that will continue to improve with AI algorithms to help break the bottleneck of segmenting and landmarking image data.
More broadly, the combination of 3D modeling, AI, and 3D printing shows the benefit of integrating multiple processes into a single workflow. While AI and 3D printing in a clinical context still presents some obstacles in terms of regulatory requirements, we are seeing this area emerge as important for industry and academics.
For now, collaboration between medical device companies, clinicians, and commercial software developers is needed to continue to refine tasks, and to leverage the experience already gained to make AI, 3D visualization, 3D printing, and other processes easier, including more dedicated labs and 3D printing hubs in hospitals.
Synopsys, Inc. is the Silicon to Software™ partner for innovative companies developing the electronic products and software applications we rely on every day. As an S&P 500 company, Synopsys has a long history of being a global leader in electronic design automation (EDA) and semiconductor IP and offers the industry’s broadest portfolio of application security testing tools and services. Whether you’re a system-on-chip (SoC) designer creating advanced semiconductors, or a software developer writing more secure, high-quality code, Synopsys has the solutions needed to deliver innovative products.
Headquartered in Cirencester, UK, Corin is a fast-growing international orthopaedic company with a vision to revolutionize orthopaedics by gaining, understanding and sharing insight at every stage of the arthroplasty experience. The unique combination of advanced technologies, shared knowledge and clinically proven implants is intended to deliver better outcomes and maximize healthcare value for patients, surgeons and healthcare providers.
About the Author:
Kerim Genc is the Business Development Manager for the Simpleware Product Group at Synopsys. He joined Simpleware in 2011 and is currently responsible for managing global sales, business development/partnerships and technical marketing content development. He received his BS and MS in biomechanics from the University of Calgary and the Pennsylvania State University respectively and completed his PhD in Biomedical Engineering at Case Western Reserve University examining countermeasures to, and the mechanics of, spaceflight induced bone loss and fracture risk.