- » View All
Design Considerations for Pharmaceutical Machine Vision
by Winn Hardin, Contributing Editor - AIA Posted 10/17/2011The flexibility and variety of machine vision solutions – smart cameras, PC-based, and turnkey systems – continues to attract growing interest from pharmaceutical industry. Machine vision technology can be found in virtually every step of the pharmaceutical discovery and manufacturing process, from automated fluorescence and other microscopy-based drug discovery systems, through pharmaceutical manufacturing, tracking, packaging and shipping.
In recent years, MachineVisionOnline.org has covered the impact of U.S. and European regulations, such as CFR 21 Part 11, which dictate how pharmaceuticals are manufactured and tracked with emphasis on manufacturing process and equipment validation and e-pedigrees that help manufacturers quickly locate defective product should on store shelves.
In this article, we’ll look at the growing number of machine vision pharmaceutical applications and design tips that can help lead to a successful system.
Tracking Remains #1
Government regulations require pharmaceutical makers to track every product down to the smallest packaged unit, such as a bottle or package of syringe needles. This means marking each product with information about where it was produced, the materials and lots used in its manufacture, any stops along the way – such as shipping to a third-party packaging company – and the final destination for the product. All this information has pushed pharmaceutical manufacturers to move from standard 1D barcodes to 2D data matrix codes And 2D data matrix codes typically mean image-based auto ID readers rather than a laser scanner.
Image based auto ID readers offer another benefit to 2D matrix compatibility: they can also read text, or perform optical character recognition/verification (OCR/OCV) for product tracking and serialization. To keep counterfeit pharmaceuticals from entering the market place, many pharmaceutical companies use UV-reflective inks, also called ‘bright stocking’ and ‘nude labeling.’ Other than radio frequency identification (RFID) tags, only machine vision systems can read UV sensitive tags.
“Optics play an important part when selecting targeted application components, such as auto ID readers, for OCR and OCV applications,” explains David Dechow, machine vision integrator and President of Aptúra Machine Vision Solutions (Lansing, Michgian). The temptation is to go with inexpensive, smaller optics. However, vignetting and roll of can lead to poor images or lower light levels at the edge of the sensor, complicating OCR/OCV applications. “A standard, security-grade 35 mm lens can see a 20% drop in illumination levels at the edge of even smaller sensors, such as 640x480 pixel array, punctuating the need for high-quality optics” adds Dechow.
Smart Cameras Gain Traction
Smart cameras, and emerging PC Cameras, offer a step up in performance and flexibility compared to traditional auto ID systems.
“Our pharmaceutical customers have requested all types of machine vision systems from smart cameras, to PC-based to turnkey systems, but smart cameras seem to be the favorite,” explains John Salls, President of Vision ICS (Woodbury, Minnesota). Machine builders and end users can spend 3 days in training and take ownership of a smart camera system. They also like the cost savings.”
Salls recently installed 13 Cognex Corporation (Natick, Massachusetts) Insight smart cameras systems on a transdermal patch manufacturing line. The patches contained powerful painkillers, and the manufacturer needed to make sure that each package contained only one patch, while checking lot codes and labeling. “If there were more than one patch in the package, and a nurse or patient opened that package, they’d be going to the hospital for sure because of the strength of the medication,” added Salls. “While we’re not seeing a lot of new lines and products, we are seeing the pharmaceutical market adopt more and more automation on their existing product lines…Lots of retrofits.”
The transdermal patches are packaged in foil wraps, which is not uncommon to pharmaceutical packaging. These highly reflective packages, along with blister packs and other metallic containers, pose unique challenges to the machine vision system. Diffuse illumination can help reduce the difference between high-intensity pixels and low spots on the same image, and avoid false rejects.
Turnkey Simplifies Compliance
For high speed, multi-camera, color and other challenging pharmaceutical applications, many clients prefer a ‘turnkey’ system because it is completely engineered to solve a specific task and can be repeated in multiple locations without have to revalidate the entire system – as long as no changes are made to either the hardware or software.
“We’re seeing that many pharmaceutical companies would much rather hire a company to develop the system rather than having staff a vision engineer for tough applications,” explains Lance Oliver, Project Engineer at VMT (Vision Machine Technic), GmbH.
Aptúra’s Dechow points out that many cost effective color cameras use Bayer filters, reducing the spatial resolution to 1/3rd of the overall sensor resolution. The situation can be further complicated by pharmaceuticals that use color packaging that closely resembles the color of the pill in a blister pack, for example, or pills with textures that can affect the color response at the sensor. High-bit depth color cameras can help improve spectral resolution while “3 chip” cameras that use one sensor for red, green, and blue channels offer maximum resolution, but both options add costs to the system.
“Optics and lighting are critical in color applications,” concludes Dechow. “Lens aberrations in the color plane can introduce significant error, while consistent light color response is required to accurately reproduce color features and shades.”
Although regulations that drive pharmaceutical companies towards automated manufacturing and quality assurance solutions have been a boon to the machine vision industry, Vision ICS’ Salls warns that CFR 21 Part 11 isn’t a black and white rule that clearly spells out what a machine vision designer has to do to make sure their customer is in compliance.
“Customers have different interpretations [of CFR 21 Part 11],” says Salls. “One customer will take the regulation very seriously, spending months of testing to try and trip up the system, while another customer will realize there is a low priority production problem with an existing system and avoid making any corrective change because of they would have to start the validation process all over again. I’m not sure how the machine vision industry can address this problem, but it would helpful to create some standards document or process that would guide integrators and our customers outlining the specific regulations that pertain to the vision systems and detailing specific steps to insure compliance.”
In the meantime, machine vision designers will continue to use the latest technology and their unique expertise to help pharmaceutical customers guarantee the quality of their products while protecting against risk and liability. While that will continue to represent a challenge to machine vision designers, the good news is that more industries – such as the food processing and some consumer products – are looking down the road, expecting additional government regulations that will require automation to solve. This can only be good for the machine vision industry, which in 2010 accounted for $75.9 million in machine vision financial transactions according to market research conducted by Paul Kellett of the Automated Imaging Association.
There are currently no comments for this article.
Leave a Comment:
All fields are required, but only your name and comment will be visible (email addresses are kept confidential). Comments are moderated and will not appear immediately. Please no link dropping, no keywords or domains as names; do not spam, and please do not advertise.