CNN Subvisible Particle Classification Biopharmaceuticals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for classifying subvisible particles in biopharmaceuticals, such as those used in sterile formulations, face challenges in accurately distinguishing between different types of particles due to their complex morphologies and textures, leading to potential inaccuracies and inefficiencies in quality control.
Innovation Solution
A particulate classification model trained as a convolutional neural network (CNN) is implemented to analyze micro flow imaging data, capable of automatically or semi-automatically identifying features like air bubbles, silicone oil droplets, and instrument artifacts, thereby reducing false positives and improving classification accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If simple image-based filters like particle aspect ratio and particle mean intensity are used for particle classification, then the classification process is simple and fast, but the classification accuracy is insufficient due to the complexity and diversity of particle morphologies
Solution Approach 1:
The patent replaces traditional mechanical/image-based filtering methods with a convolutional neural network (CNN) system. The CNN automatically learns complex morphological features from particle images, substituting manual feature engineering with automated deep learning-based feature extraction and classification, thereby achieving both high accuracy and efficiency
Solution Approach 2:
The patent transforms the classification approach by changing from fixed threshold-based parameters (aspect ratio, mean intensity) to adaptive learned parameters through CNN training. The network learns optimal feature representations and decision boundaries from training data, enabling accurate classification of diverse particle morphologies while maintaining computational efficiency
2Measurement precision
If a comprehensive particulate classification model is trained to accurately classify all types of subvisible particles, then classification accuracy improves, but computational resources and model size increase
Solution Approach 1:
The patent segments the classification task into two stages: first, a detection model identifies and localizes particles in MFI images; second, a classification model categorizes detected particles. This segmentation allows each model to be optimized for its specific function, reducing overall computational complexity while maintaining comprehensive classification accuracy
Solution Approach 2:
The patent performs preliminary particle detection and localization before classification. By pre-identifying particle regions of interest, the system avoids processing entire images for classification, significantly reducing computational resources and model size requirements while maintaining accurate classification of all particle types
Data Source
AI summary
An analysis system implements a particulate classification model trained as a convolutional neural network to assess sterile formulation quality. The system obtains micro flow imaging (MFI) data for a sterile formulation drug product, wherein the MFI data includes an image depicting a plurality of sub-visible particulates in the sterile formulation drug product. The system detects a plurality of features in the image, wherein each feature is bounded by a bounding box. The system performs preprocessing to determine and remove one or more of features to be artifacts based on characteristics of the features, wherein the remaining features are determined to be sub-visible particulates. The system applies a particulate classification model to each sub-visible particulate to determine a particulate classification label. The system determines a quality metric for the sterile formulation drug product based on a count of sub-visible particulates in each particulate classification label.