CDER Researches Convolutional Neural Networks
CDER researchers are working on applying artificial intelligence/machine learning in the form of convolutional neural networks (CNNs) to enable the processing of large collections of images with high efficiency and accuracy by distinguishing complex “textual features” that are not readily delineated with existing image processing software. An online report describes a case study in which CDER researchers and collaborators at Ursa Analytics, the University of Colorado, and the National Institute of Standards and Technology implemented CNNs to further the understanding of certain product quality attributes for a model therapeutic protein formulation.
“The study demonstrated the advances in microscopy analyses and a tool to monitor changes in product quality attributes can be achieved through CNN-based approaches, and the study’s findings may be applicable to evaluation of a variety of biopharmaceutical drug products,” the report says.
CDER says the methodology applied in the study is applicable to a range of products in pharmaceuticals and biopharmaceuticals to monitor changes in product attributes such as particles/aggregates during manufacturing. “The analytical procedure (flow microscopy combined with CNN image analysis) explored in the CDER research study can detect small shifts in protein aggregate populations due to stresses resulting from unknown process upsets providing potential new strategies for monitoring product quality attributes,” the report says. “In addition, use of a reference standard such as ethylene tetrafluoroethylene that is stable over time and possesses optical properties similar to those of protein aggregates is useful for validating and evaluating the robustness of the analytical procedure.”