Abstract
With machine learning (ML) we see the potential to better harness the intelligence and decision-making abilities of human inspectors performing manual visual inspection (MVI) and apply this to automated visual inspection (AVI) with the inherent improvements in throughput and consistency. This paper is intended to capture current experience with this new technology and provides points to consider (PtC) for successful application to AVI of injectable drug products. The technology is available today for such AVI applications. Machine vision companies have integrated ML as an additional visual inspection tool with minimal upgrades to existing hardware. Studies have demonstrated superior results in defect detection and reduction in false rejects, when compared with conventional inspection tools. ML implementation does not require modifications to current AVI qualification strategies. The utilization of this technology for AVI will accelerate recipe development using faster computers rather than by direct human configuration and coding of vision tools. By freezing the model developed with AI tools and subjecting it to current validation strategies, good assurance of reliable performance in the production environment can be achieved.
- Automated Inspection
- Deep Learning
- Injectable Drug
- Machine Learning
- Supervised Learning
- Visual Inspection, Unsupervised Learning Image Labelling Neural Network, Inspection Qualification
- Received September 10, 2022.
- Accepted June 7, 2023.
- Copyright © 2023, Parenteral Drug Association
PDA members receive access to all articles published in the current year and previous volume year. Institutional subscribers received access to all content. Log in below to receive access to this article if you are either of these.
If you are neither or you are a PDA member trying to access an article outside of your membership license, then you must purchase access to this article (below). If you do not have a username or password for JPST, you will be required to create an account prior to purchasing.
Full issue PDFs are for PDA members only.
Note to pda.org users
The PDA and PDA bookstore websites (www.pda.org and www.pda.org/bookstore) are separate websites from the PDA JPST website. When you first join PDA, your initial UserID and Password are sent to HighWirePress to create your PDA JPST account. Subsequent UserrID and Password changes required at the PDA websites will not pass on to PDA JPST and vice versa. If you forget your PDA JPST UserID and/or Password, you can request help to retrieve UserID and reset Password below.