Application of Deep Learning Convolutional Neural Networks for Internal Tablet Defect Detection: High Accuracy,...

Area of Science:

• Pharmaceutical Manufacturing
• Quality Control
• Artificial Intelligence in Drug Development

Background:

• Internal tablet cracking is a critical defect in oral formulations, leading to batch failures and increased costs.
• Traditional X-ray computed tomography (XRCT) analysis for internal cracks is manual, subjective, and time-consuming.
• Current methods lack efficiency and consistency for large-scale industrial application.

Purpose of the Study:

• To develop an automated deep learning program for analyzing internal tablet cracks using XRCT data.
• To enhance the accuracy, efficiency, and objectivity of tablet defect detection in pharmaceutical manufacturing.
• To assess the adaptability and industrial applicability of the developed deep learning tool.

Main Methods:

• Development of a deep learning convolutional neural network (CNN) program for automated XRCT image analysis.
• Quantification of internal tablet cracks and assessment of detection accuracy.
• Evaluation of the program's throughput and adaptability to different pharmaceutical products.

Main Results:

• The deep learning program achieved an average accuracy of 94% in detecting internal tablet cracks.
• The automated tool demonstrated high throughput, capable of analyzing hundreds of tablets.
• The analysis program showed adaptability to various tablet types and packaging.

Conclusions:

• Deep learning-based automated XRCT analysis provides a robust solution for detecting internal tablet cracks.
• The developed tool significantly improves the efficiency and consistency of quality control in pharmaceutical manufacturing.
• The successful implementation into the industrial workflow highlights the potential of AI in pharmaceutical quality assurance.