Machine vision focuses on quality

Quality control takes on a whole new meaning in the world of medical device manufacturing. While companies in most other industries merely use high quality to gain a competitive edge, the makers of medical products typically need to ensure absolute quality to bring their products to market.

A leading medical manufacturer's inspection cycle takes approximately 90 msec per device and about a half-second for an entire row of six cholesterol home-screening kits using the Checkpoint 800 system from Cognex Corp.

That quality mandate runs deep. Government regulations, such as the Food and Drug Administration's General Manufacturing Practice (GMP) guidelines in the US, usually make quality assurance an ongoing requirement. In many situations, the production processes themselves often must meet the FDA's exacting certification standards. Then, of course, there is the problem of product liability. Even a single flaw or faulty component can bring on disastrous lawsuits.

Thus, medical device makers face a two-fold quality challenge. First, they must achieve zero-defects in their production processes. Then, they often must be able to provide documented proof that verifies it. Many companies have found that using machine vision for automated inspection and machine guidance is the best overall solution to this challenge.

Bringing devices to market

Most automation tools allow manufacturers to improve quality or efficiency in some way. But every so often, one of those tools also proves to be crucial to the commercial viability of an entirely new product category. One leading medical device company relied upon such a mission-critical technology to bring cholesterol home-screening kits to market.

The enabling technology that makes it possible to mass-produce the kits to the required quality standard is machine vision.

After nearly five years of research and development, the company received FDA clearance to market the product. The design is a model of simplicity; a user simply pricks a finger with a tiny lancet and places a drop of blood into a small sample well at one end of the device. The blood is automatically absorbed into a series of four tiny pads where it reacts with the biochemical compounds. When the reaction process is complete, the device produces a cholesterol reading in a capillary channel running along its length.

Engineering a high speed production process for the numerous lightweight components of the device required exacting tolerances and control over several assembly operations in a single manufacturing line. The company used custom thermoforming equipment to produce the thin, plastic device body in webs of six units across. Then, assembly mechanisms precisely placed the reagent pads within the small well of each device.

To meet the stringent quality requirements, the company needed to verify zero defects through automated inspections of each device during manufacturing operations. From the start, it was evident that this would require powerful pattern recognition capabilities available only with advanced machine vision technology.

The company turned to a system integrator to help deal with the complexity and to engineer the most effective solution. The project team then set about the task of choosing the right machine vision platform. Ultimately, the technical demands and programming requirements of the application led to the Checkpoint 800 machine vision system manufactured by Cognex Corp, Natick, MA.

The Checkpoint system combines industrial caliber machine vision capabilities with the usability of Microsoft Windows. The main advantage of this platform is that it enabled the company to gain access to vision processing hardware and image analysis software tools through a streamlined application development environment. Checkpoint enabled comparatively rapid application deployment through a Microsoft Windows-based "point and click" programming environment. From start to finish, the system integrator completed the application development phase of the project in just over six weeks. This reduced the company's inspection cycle to approximately 90 msec per device and about a half-second for an entire row of six.

Machine vision fundamentals

Few industries have as urgent a need as medical. The primary goal of most machine vision systems is to improve productivity and quality in the manufacturing process. In a typical application, a sensor detects the presence of the part and signals the vision system to activate a video camera, positioned above or to the side of the inspection point, to capture an image of a part or subassembly and send it to the machine vision processor.

While companies in most other industries merely use high quality to gain a competitive edge, the makers of medical products often need to ensure absolute quality to bring their products to market. Above are two examples of how machine vision allows medical device manufacturers to make products with zero defect results. Pictured from left to right are a pacemaker and an IV Bag.

To work with other types of equipment and automation systems on the plant floor, machine vision now comes in many industrial form factors. They include: standalone, rack -mounted, and PC plug-in cards for PCI, Compact PCI, and VME bus architectures. Using a combination of hardware and machine vision software, a vision system analyzes the image, usually in a fraction of a second. In this stage, it might determine where the part is located, analyze its orientation, measure critical dimensions, or verify correct assembly. At the completion of each image analysis, the machine vision system communicates the results to other factory equipment--such as programmable logic controllers (PLCs), robots, or data collection computers.

Advanced manufacturing techniques and processes now place extraordinary demands on automation equipment. To be fully functional on the plant floor, a machine vision system needs to be precise, reliable, and repeatable at high throughput rates. This means that the system must have powerful gray-scale processing and analysis capabilities. The best new generation vision systems typically have the following features and capabilities:

• Gray scale processing which digitizes and analyzes images in 256 shades of gray.
• A set of tools for locating parts and features--often the first step in solving most vision problems.
• Fast processing speed--typically as quick as 5 to 10 msec per inspection to meet the demands of high speed production lines.
• Flexibility to add capabilities to existing applications and address new applications that may arise.

Powerful application development software is also one of the hallmarks of advanced machine vision systems. Traditionally, the development tools in sophisticated systems have required expertise in a professional-level computer programming language such as "C." The new generation of systems has changed that with simpler Windows-based "point and click" graphical environments that make it easier for plantfloor users to develop and deploy vision applications.

Vision improves quality

The vast majority of applications in the medical device field are for machine guidance or inspections. In quality control inspection applications, the vision system determines whether parts or subassemblies are acceptable or defective and then directs motion control equipment to reject or accept them. Machine guidance applications use vision systems to improve the accuracy and speed of assembly robots and automated material handling equipment.

Although they can vary in many ways, the applications are usually in one of several general categories.

Robotic guidance--Traditionally used in heavy manufacturing applications such as those in the automotive industry, robots are now working their way into the medical device field as well. With an inherent ability to handle a range of highly repetitive tasks, robotics are among the most flexible forms of automation. Machine vision enhances that flexibility by enabling a robot to see the part or subassembly on which it is working. In most applications, machine vision systems provide real-time and live feedback to guide robots as they go through programmed sequences of operation. To perform this level of machine guidance, a vision system usually locates parts for the robot to pick up, identifies the correct location at which to place the parts, and sends this information to the robot for the operation.

Machine vision can also be used to verify that laser gradation marks on devices, such as this syringe, are correct and legible.

Dimensional gaging--For many medical devices, quality is a matter of small dimensions and details. The new generation of machine vision systems excels at ensuring that those measurements are correct. These dimensional gaging applications typically encompass lines, angles, arcs, and diameters.

Assembly verification--Medical devices are more sophisticated than ever, and as a result, the subassemblies in them are often extremely complex. This imposes a difficult task on manufacturers--making sure that the subassemblies include all parts, and that each of those parts are in place correctly. This is where the new generation of machine vision can really prove its worth. By training the systems to find detailed patterns and shapes, users program the systems to compare images against templates of correctly made subassemblies. Almost without exception, the results are far more accurate and reliable than would be possible with any other quality control method--manual or automated.

Gaging of medical devices, such as these surgical staples, provides the required FDA and GMP zero defect results medical device manufacturers are required to achieve.

Flaw detection--In a medical device, material flaws are seldom just cosmetic; they can affect product performance. The concept of zero defects means making absolutely sure that customers never receive a flawed item. Thus, this type of quality inspection has become one of the most common machine vision applications in the industry. Vision systems are used to find missing material, chips, scratches, dents, misplaced markings, and other flaws. These inspections produce a secondary benefit by enabling manufacturers to eliminate defective pieces before wasting additional material and production time on them, thus controlling costs and minimizing material waste.

Print verification--Using various optical character verification (OCV) methods, machine vision systems inspect parts, components, and products to make sure that they are labeled and marked correctly. In the real world of the production line, this can be more complex than it might initially seem. The systems may have to compensate for variations in character density, inking, and shape, as well as the secondary effects of laser etching, stamping, and engraving. For just that purpose, the best machine vision technology allows users to adjust the application quickly with little or no delay in the production process.

Code reading--The FDA now mandates full traceability for many devices according to vendor and production lot. To meet this requirement, some companies are laser-etching OCRA and 2D data matrix codes on their products, and using machine vision to read the codes. They then can take advantage of the information for tracking, verification, and statistical quality control.

For more information from Cognex Corp, Natick, MA, circle 411; http://www.cognex.com

Originally published in the October 1998 issue
of Tooling & Production.
Please Note: some pictures or diagrams are only available through the printed media.