Skip To The Main Content

Chapter Covers New Developments in Visual Inspection

Visual inspection is a major topic of interest to PDA members around the world. On Nov. 22, the PDA Ireland Chapter held a workshop in Kildare focused on "Visual Inspection — Requirements, Practical Implementation and Future Technologies." 

The major critical takeaways from the event cover manual inspections, automated inspections, regulatory developments, achieving zero defects for visible particles, new technologies and difficult-to-inspect products.

Manual Inspection

To meet the requirements for inspector training and qualification, separate defect sets for training, size sensitivity and qualification need to be created. The manual baseline should consider inspection conditions (e.g., station, lighting), inspector technique (e.g., pacing), individual inspector variability and person-to-person variability. When multiple container types, sizes and fill levels are used in a manufacturing facility, it may be possible to develop a bracketing approach to operational challenge sets. Acceptance criteria for manual inspection qualification criteria is strongly influenced by qualification set design.

The first step in developing a validation challenge set is to complete a risk assessment for the entire process, including component manufacturing—container, stoppers, product formulation, container filling and container handling, for example. A well-designed challenge set will ensure that automated processes are robust, and that defect detection is reliable. Including hard-to-detect defects in sets results in lower acceptance criteria.

Lower acceptance criteria should not be confused with lower process capability, however. Clear and detailed descriptions of created defects need to be included in procedures to ensure consistency over time and the ability to replace defects (e.g., written descriptions, pictures). If the created defect is not a direct copy of the natural defect, performing studies that demonstrate comparable performance for visual inspection should be considered. Having a specialized group or specially trained individuals devise sets would produce clear benefits.

Automated Inspection

With the advances in technology, many companies are exploring automated inspection. To implement an automated system, though, first requires developing a high level of competence with manual inspection. Having a thorough understanding of the product and its inherent variability is required before it can be migrated from manual to automated visual inspection. For example, a company would need to develop in-house expertise in machine vision, or it would have to rely on vender support and availability.

Should it come to selecting a vendor, having a formal process in place for commercial bid analysis allows for visual and transparent decision-making. It would also align with the desired values and culture of lean, enabling teams to emphasize a project’s value to its stakeholders. One speaker relayed how one company took to the road, delivering samples door-to-door to mitigate the impact of shipment in defect samples. To limit potential false rejects due to container variability, Schott created vials with dimensions at limit of specifications (maximum, minimum and nominal) for heel radius, shoulder, push-up and bottom thickness.

Qualification maintenance was also discussed as an alternative to requalification with defect panels in an annual review process. This would only be valid if a robust inspection lifecycle process, per USP <1790> Visual Inspection of Injections, was followed (1). Trending of rejects over time is key: If an automated inspection method is shown to be better than the manual method, does this call into question the use of the manual method for reinspection?

Also having the capability to conduct offline investigations, that is, a mirror copy of a vision system on an offline laptop, was highly recommended.

Regulatory Developments

Audience in attendance to the in a ballroom at their tablesDefects remain one of the Top 10 reasons for drug product recalls, with more than 200 incidents since 2008. From the regulators’ perspective, if you can see it, you should control it!

The quality risk management (QRM)/lifecycle approach is the key to particulate matter control—what presents a risk to patients?

Not only equipment and processes, but the human factor should be considered. Limitations and stressors should be addressed as part of the operator qualification program, which should also include fatigue studies. And QRM processes should be integrated as part of operator training.

Achieving Zero Defects for Visible Particles in Parenterals

When it comes to the topic of achieving zero defects for visible particles, PDA is involved in numerous activities. One study, initiated in 2017 by an industry cross-functional group (co-sponsored by PDA and the Pharmaceutical Manufacturing Forum), that addresses particle sources in pharmaceutical products is ongoing. Another task force is focusing on visible particles in ready-to-fill, use-and-sterilize components, glass and elastomer components and secondary packaging associated with packaging components. Outcomes from some of the studies to date have been published in the PDA JPST and publication of future results is anticipated.

New Technologies

The industry is rapidly evolving with respect to its manufacturing processes and advanced therapies. Inspection technologies need to evolve in parallel.

Interest in applying deep learning and artificial intelligence (AI) technologies to visual inspection is growing. Factories of the future will likely feature a vision robot unit that incorporates deep learning and AI and is integrated with factory systems. Deep learning technology will be reactive, that is, the inspection reacts to product characteristics during one batch and adapts. Continuous learning ensues from production and from operators’ retraining of new defects and false rejects reduced by retraining false positives.

Difficult-to-Inspect Products

Many of those attending this meeting work primarily with lyophilized products. For these products, particles are revealed only at the surface. Organizations have developed a consistent approach with respect to destructive testing methodology (USP <788> Particulate Matter in Injections) and sample sizes used for lyophilized products (2). Reconstitution is typically performed in a controlled environment using a terminally filtered diluent and mixed to obtain a clear solution. While discussed, the filtration method is not used by participants; instead, the reconstituted and filtered product is examined by microscope.

While workshop participants did not have experience with other difficult-to-inspect products, John Shabushnig highlighted PDA Technical Report No. 79: Particulate Matter Control in Difficult-to-Inspect Parenterals (3). This technical report describes best practices for difficult-to-inspect parenteral product lifecycle management, destructive testing, and trending to supplement portions of the guidance given in USP <1790>.

During the Q&A portions of the meeting, participants shared their experiences with foreign regulators’ expectations of visual inspection. Two participants reported that Russian inspectors had expressed a requirement for a Grade D background for inspections. Chinese inspectors envisage inspection taking place in a quiet, dark room with no light, apart from the booth lighting; and, in general, they prefer automated inspection. Japanese inspectors tend to have high expectations and requirements for cosmetic defects.

All in all, the meeting proved a success! The presentations were insightful and the discussion engaging.


  1. U.S. Pharmacopeial Convention. General Chapter <1790> Visual Inspection of Injections. In USP 41-NF 36, USP: Rockville, Md., 2018.
  2. U.S. Pharmacopeial Convention. General Chapter <788> Particulate Matter in Injections. In USP 41–NF 36, USP: Rockville, Md., 2018.
  3. Parenteral Drug Association. Technical Report No. 79: Particulate Matter Control in Difficult-to-Inspect Parenterals; PDA: Bethesda, Md., 2018.

PDA Members Save Substantially