Mimicking the human decision-making process is challenging. Especially, many process\ncontrol situations during the manufacturing of pharmaceuticals are based on visual observations\nand related experience-based actions. The aim of the present work was to investigate the use of\nimage analysis to classify the quality of coated tablets. Tablets with an increasing amount of coating\nsolution were imaged by fast scanning using a conventional office scanner. A segmentation routine\nwas implemented to the images, allowing the extraction of numeric image-based information from\nindividual tablets. The image preprocessing was performed prior to utilization of four different\nclassification techniques for the individual tablet images. The support vector machine (SVM)\ntechnique performed superior compared to a convolutional neural network (CNN) in relation to\ncomputational time, and this approach was also slightly better at classifying the tablets correctly.\nThe fastest multivariate method was partial least squares (PLS) regression, but this method was\nhampered by the inferior classification accuracy of the tablets. Finally, it was possible to create a\nnumerical threshold classification model with an accuracy comparable to the SVM approach, so it is\nevident that there exist multiple valid options for classifying coated tablets.
Loading....