Menu
Top 5 Global Medical Device Manufacturer Implements & Validates Computer Vision Solution in Record Speed
Improving the reliability, speed, and quality of their operations
CASE STUDY
A global leader in medical technology, offering medical devices and therapies across over 100 countries.
During the manufacturing and assembling operations for implantable medical devices, quality is of utmost importance. If a company fails to detect defective parts, the cost can be very high. Medical device manufacturers must take action to analyze the root cause and implement corrective procedures to avoid an escape in the future.
The client’s multi-stage defect inspection process was manual and required several inspectors. First stage inspectors were less trained and sometimes let defects slip through the cracks. In addition to the inspection process being unreliable and not reproducible, it was costly and training intensive.
The client sought a computer vision solution that would identify defects and allow for continuous improvement through machine learning to improve the reliability, speed, and accuracy of their manufacturing processes. An inhibitor to innovation and implementing transformative technology is regulatory compliance – specifically upfront and ongoing validation. Typical validation times for a solution like the one selected could take 2-3 months, and that’s if all went smoothly. The client needed a specialized partner who not only understood validation, but also vision automation software that could shorten the time to solution implementation. Since cloud-based software solutions deliver upgrades on a continual basis, the client also needed help with ongoing validation in a timely and cost-effective manner.
The combination of LandingLens (Landing AI’s computer vision platform) and Verista’s validation and vision automation expertise, helped the client achieve their goal to implement and validate an AI vision inspection software that identifies defects and continually improves itself through machine learning in record time and at a cost-effective price.
Landing AI’s platform includes a standardized workflow with traceability built in so it’s easier to document and validate. This, along with Verista’s validation package, allows the customer to scale more quickly and effectively, reducing validation from months to weeks. The combination can be easily replicated from line to line and plant to plant.
Verista was selected as the validation partner for Landing AI due to our expertise in manufacturing equipment validation as well as validation of cloud-based solution platforms, a rare and unique find in the marketplace. The LandingLens system was validated in two parts: the platform and the model. The LandingLens platform is used to collect and tag images into a data set, build and train image models, as well as deploy the models for production use. The model is used to interpret the images collected during the manufacturing process. By qualifying the platform separately, the system can be leveraged by multiple production lines or facilities within the organization while the models are independently validated for each use case.
Verista’s unique platform validation package includes functional requirement specifications, installation verification and operation qualification protocol, and requirements trace matrix. This package allows the client to execute the protocol using a standard set of images to give a high degree of confidence in the functionality of the platform without requiring client specific process images.
Deployment and Ongoing Validation
The powerful combination of Landing AI’s LandingLens and Verista’s validation capabilities to meet FDA compliance reduced the initial deployment validation time from 3+ months to 1-2 weeks. This allowed the client to quickly implement the solution in their production environment.
In addition, since Landing AI is a cloud-based software, they deliver continuous updates and features, requiring the client to revalidate the software in the manufacturing environment. Verista can step in and ensure that the validated state is fully maintained. Furthermore, when the client makes a manufacturing change, i.e., a change to a product or device, Verista can quickly retrain and revalidate the AI model.
CLIENT BENEFITS
In addition to improving the reliability, speed, and accuracy of their manufacturing processes, the client experienced the following benefits:
✓ Improved quality, reduced product error escapes, and greater insights through the use of automation and Landing AI’s vision inspection technology
✓ Higher quality yields and increased speed to market through the real-time capture and AI analysis of defects
✓ Increased operational efficiency through the reduction of CAPAs
✓ Cost-effective price: 2X less due to vision AI and validation expertise
✓ Cloud-based technology is easily scalable
✓ Faster and easier validation through the use of Verista’s comprehensive validation package and experti
