Wavelet image decomposition for characterization of freeze-dried pharmaceutical product structures

This paper investigates the use of a wavelet image decomposition applied to electron microscope images in order to estimate the mass transfer coefficient of pharmaceutical cakes obtained by freeze-drying. The structure analysis of dried cakes obtained by means of a free-drying process, is a basic step for tuning the process conditions and for monitoring the quality of the dried product. The product structure and specifically its porosity affects the drying duration as it defines the resistance to the vapor flow during the ice sublimation. This parameter is becoming quite important as it is fundamental for modeling of the freeze-drying process and thus for an optimal design of the freeze-drying cycle. The direct measurement of this parameter is quite complex thus new simple approaches are being developed for its non-invasive estimation. This paper discusses the possibility of processing SEM images of the dried cake to analyze its morphology and to estimate the mass transfer coefficient. This approach has already been followed by processing the images via a 2D-FFT, here a faster solution based on the image wavelet decomposition followed by a non-linear processing based on an artificial neural network is described and the results are compared with the one obtained by the traditional direct mass transfer coefficient measurement.