OR WAIT null SECS
The industry is taking steps to automate the final product inspection process for complex therapeutics.
Prior to a pharmaceutical product making its way into the supply chain, companies must perform extensive inspections to check for any potential defects and ensure the utmost quality. The market is leaning toward more complex therapeutics, which are sensitive to a variety of environmental factors and expensive to develop, and continuous, automated processes. These developments are making efficient inspection of the finished product ever more pressing.
Traditionally, finished product inspection is still widely performed manually; however, automating the process, either fully or partially, can allow for increased efficiencies in both time and cost (1). In fact, according to market research, the fully automated inspection machines segment, which comprised 42.5% of the market share in 2022, is predicted to experience significant growth between 2023 and 2032 (2).
To learn more about the evolution of finished product inspection, the types of automated technologies available, regulations, challenges, and future trends in the field, Pharmaceutical Technology® spoke with Wolfram Schindler, global product manager for Inspection Technology at Syntegon.
PharmTech: Could you provide a brief overview on the evolution of finished product inspection?
Schindler (Syntegon): The first electronic particle inspection system was developed by the pharmaceutical company Eisai Co., Ltd together with the Technical University of Tokyo and patented in 1970. The ‘static division’ or ‘SD’ sensor technology, now belonging to Syntegon, is based on transmission light and a diode array to detect moving foreign particles in liquids but neglects static particles on the container surface. Five years later, the first fully automated inspection machine AIM275 was introduced to replace manual inspection at a higher output. Along with technical progress in machinery and imaging technology, today automated visual inspection machines can reach speeds of 600/min and more—two orders of magnitudes higher than manual inspection. Automated inspection is following the general trend from ampoules, vials, and pre-filled syringes to even more complex container types. This evolution poses challenges in terms of mechanical handling and vision capability. Since the early 1990s, camera technology has extended inspection capability towards non-moving particles and cosmetic defects like cracks in the glass or container-specific defects (ampoule tip shape, vial crimp, syringe flange, and stopper side, etc.).
Manual inspection is still widely used, especially for difficult-to-inspect parenteral products. For 100% batch inspection in general, however, choosing between manual or fully automated inspection is rather a question of batch size than technical capability. Finally, semi-automated inspection machines may bridge the gap between manual and fully automated inspection systems for several applications, offering a good balance between cost, speed, and inspection performance. Sometimes inspection approaches are combined. In a customer case involving molded glass, the human eye first identifies glass defects. The downstream automated inspection machine, which in this case would lead to high false reject rates for cosmetic defects in the inhomogeneous glass, is activated only for moving particle detection.
Besides visual inspection, physical methods for automated container closure integrity testing (CCIT) have been available on the market for decades, including high voltage leak detection (HVLD), the vacuum/pressure method and, for about 20 years, headspace analysis.
PharmTech: What are the different types of automated finished product inspection technologies?
Schindler (Syntegon): The most common machine concept is pretty similar to the 50-year-old patent—a turret with inspection stations. Up to this day, an advanced version of the SD sensor technology is available, which various customers use for moving particle detection.
However, nowadays visual inspection is mainly based on matrix camera stations. Technical specifications of such stations may significantly differ in terms of frame rate, resolution, or dynamic range. Line scan cameras are another option (e.g., for 360° views without distortions). Depending on the requirements, one or more imaging technologies can be combined on the same machine. Particularly for lyophilized products, NIR [near infrared] spectroscopy or X-ray imaging may serve as a supplementary technology but are not that common.
Regarding CCIT, HVLD is widely used for leak detection of containers filled with conductive liquid. Another option for leak detection is the vacuum or pressure decay method. Laser-based head space analysis (HSA) is particularly suitable to identify both active and temporary leaks in lyophilized product vials. Beyond pure leak indications, the HSA measurement values such as total pressure, oxygen, moisture, or CO2 are important quality attributes per se.
PharmTech: Are there any specific current trends impacting automated finished product inspection or any future trends that you could highlight?
Schindler (Syntegon): Finished product inspection should identify all relevant defective units from a quality standpoint but keep the false reject rate as low as possible from an economic and ethical perspective (e.g., think of blood plasma products). This ever-lasting key question is not trivial. Automated visual inspection and CCIT is best solved, according to Syntegon, through strong collaboration between client and vendor.
Difficult-to-inspect parenteral products, such as lyophilized products, are currently being addressed with novel technology. The challenge with lyophilized products is based on product characteristics like opacity, which makes it hard to detect particles as they do not appear on the products’ surface. Moreover, because lyophilized cakes often are not homogeneous but show cracks and rough surfaces, the distinction between particles and shadows is difficult. Product splashes on the containers’ wall also need to be clearly distinguished from particles or glass defects.
Deep learning (DL) approaches using artificial neural networks (ANN) are on the rise and will significantly improve within the next years with each new data set. DL holds the potential of further increasing detection rates and decreasing the number of false rejects compared to standard image processing tools, particularly for products with high variability among good samples (like lyophilized products).
For low outputs, robotic vision solutions together with artificial intelligence (AI) open a new path to inspect complex product-package combinations where classic automated approaches are not sufficient. When high machine speeds are required, accurate timing of cameras and lights, image processing time, and sensor signal accuracy are areas that need to be continuously optimized. Just like other advanced vision tools, AI will be increasingly considered and is currently undergoing feasibility studies in a growing number of projects.
The adequate use of the equipment is equally important, including a firm method development and validation strategy. The compilation of adequate test kits, as well as regular performance checks and re-calibration plans are vital for reliable detection processes.
In the future, there will be continuous expansion of inspection machines portfolios that combine visual and CCIT technologies at increased speed. New systems will incorporate advanced vision tools, and some CCIT sensor technologies are currently being developed further to be integrated into those solutions.
PharmTech: How are regulatory bodies adapting to automation of finished product inspection? Are there any new guidelines that companies should be aware of?
Schindler (Syntegon): There has recently been an increasing awareness for calibration topics and a data-driven, scientific approach. Machines therefore include tools for trending and the recording of data, a dedicated sampling channel or special test modes to assist the users.
It is notable that the new EU GMP [European Union good manufacturing practice] Annex 1 (effective since August 2022) addresses visual inspection and CCIT with doubled text length compared to the 2008 version and highlights the importance of process knowledge. The guidance also stresses the need to generate a defect library, to categorize defects and to trend defect types and quantities.
Moreover, Annex 1 clearly advises to verify the equipment performance using representative defects before the start of inspection and during the batch. Finally, according to Annex 1, visual inspection alone is not seen as an acceptable integrity test method, which leads to the conclusion that it has to be combined with dedicated CCIT methods for that purpose.
Felicity Thomas is the European/senior editor for Pharmaceutical Technology Group.
Vol. 47, No. 8
When referring to this article, please cite it as Thomas, F. Progressing Finished Product Inspection Through Automation. Pharmaceutical Technology, 2023, 47 (8) 28–30.