Artificial intelligence-aided endoscopic in-line particle size analysis during the pellet layering process

Area of Science:

• Pharmaceutical Technology
• Process Analytical Technology
• Artificial Intelligence in Manufacturing

Background:

• Pellet layering is crucial for drug content uniformity.
• Accurate particle size monitoring is essential for ensuring drug product safety and quality.
• Existing methods for particle size analysis may not be suitable for real-time in-line monitoring.

Purpose of the Study:

• To develop an artificial intelligence-based machine vision system for in-line particle size analysis during pellet layering.
• To enable real-time monitoring of pellet size and layer uniformity.
• To ensure timely intervention for out-of-spec products.

Main Methods:

• Developed a machine vision system using a rigid endoscope, light source, and high-speed camera.
• Employed a convolutional neural network-based instance segmentation algorithm for accurate particle detection in dense flow.
• Trained and validated the system using pellet cores of varying sizes (250-850 μm) and tested in-line during large-scale drug layering.

Main Results:

• The AI system accurately determined pellet size in real time, even with dense particle flow.
• Real-time particle size data correlated well with reference methods.
• Demonstrated the system's capability for in-line process monitoring during drug layering.

Conclusions:

• The developed machine vision system is a feasible Process Analytical Technology (PAT) tool for in-line monitoring of pellet size.
• This AI-based approach enhances control over the pellet layering process, improving product safety and quality.
• Real-time particle size analysis facilitates timely interventions, optimizing pharmaceutical manufacturing.