A micro-XRT image analysis and machine learning methodology for the characterisation of multi-particulate capsule...

Area of Science:

• Pharmaceutical Sciences
• Materials Science
• Image Analysis

Background:

• Pharmaceutical multi-particulate systems require precise structural characterization for quality control.
• Traditional methods may not fully capture the complex 3D structure of individual pellets within a formulation.

Purpose of the Study:

• To demonstrate the application of X-ray microtomography for quantitative structural analysis of pharmaceutical pellets.
• To develop a machine learning model for detecting broken pellets using extracted structural features.

Main Methods:

• X-ray microtomography was used for 3D structural analysis of ibuprofen pellets within commercial capsules.
• Marker-supported watershed transformation enabled reliable pellet segmentation and individual 3D analysis.
• Over 200 quantitative features (size, shape, porosity, surface, orientation) were extracted.
• ReliefF feature selection and Support Vector Machine algorithms were employed for model development.

Main Results:

• A classification model accurately identified broken pellets with >99.55% accuracy and 86.20% precision.
• The method successfully characterized individual pellet structures from a population of ~300 pellets per capsule.
• Analysis of over 200 features provided comprehensive structural insights.

Conclusions:

• X-ray microtomography offers a powerful tool for detailed quantitative structural analysis of pharmaceutical pellets.
• Advanced data analysis, including machine learning, can effectively detect particle defects, improving quality control.
• This approach has significant potential to optimize pharmaceutical manufacturing processes and accelerate development.