The present study was aimed to the development and characterization of valsartan-containing buccal films with an introduction to a novel technique of image analysis. Visual inspection of the film provided a wealth of information that was difficult to quantify objectively. The obtained images of the films observed under the microscope were embedded in a convolutional neural network (CNN). The results were clustered according to their visual quality and on the basis of data distances. Image analysis proved to be a promising method to characterize buccal films appearance and their visual properties. The differential behavior of film composition was investigated using a reduced combinatorial experimental design. Formulation properties such as dissolution rate, moisture content, valsartan particle size distribution, film thickness, and drug assay were evaluated. In addition, more advanced methods such as Raman microscopy and image analysis were used to characterize the developed product in more detail. The results of dissolution tests using four different dissolution apparatuses showed a significant difference between the formulations containing the active ingredient in different polymorphic states. The dynamic contact angle of a water droplet on the surface of the films was measured, which correlated well with the dissolution times at 80% of the released drug (t80).
Keywords: buccal films; convolutional neural networks; dissolution testing; image analytics; raman microscopy; valsartan.
© 2023. The Author(s), under exclusive licence to American Association of Pharmaceutical Scientists.