Compendial methods for the analysis of subvisible particles in protein drugs are inadequate. New techniques, like flow-imaging microscopy (FIM), show promise but require subjective, qualitative manual identification. These manual methods ignore relevant structural or morphological information embedded in the images. Applying a convolutional neural network (ConvNet or CNN) delivers objective, analytical capability. And, leverages the embedded morphologic properties contained in FIM images. CNN can replace the subjective manual analysis with automated quantitative data on protein aggregates.
The author discusses the use of CNN to leverage FIM images to analyze aggregates in monoclonal antibody (mAb) drugs. The results demonstrate that CNN can identify and differentiate protein aggregates based on their composition and the stresses imposed to create the aggregate.
This paper was awarded the Ebert Prize by the American Pharmacists Association (APhA)Academy of Pharmaceutical Research and Science in 2019. The Ebert Prize is presented annually by APhA to the authors of an article published in the Journal of Pharmaceutical Sciences (JPharmSci™) “which describes particularly new, original, and novel findings that have a high probability of significantly impacting the pharmaceutical sciences. You can learn more about the Ebert Prize here.
The full technical paper is embedded below.Calderon-et-al-Deep-CNN-Analysis-of-FIM-Data-to-Classify-Subvis-Particles-in-Protein-Formul