Unsupervised Deep Learning Approach for Characterizing Fractality in Dried Drop Patterns of Differently Mixed Viscum album Preparations

Abstract

This paper presents a novel unsupervised deep learning methodology for the analysis of self-assembled structures formed in evaporating droplets. The proposed approach focuses on clustering these structures based on their texture similarity to characterize three different mixing procedures (turbulent, laminar, and diffusion-based) applied to produce Viscum album Quercus 10−3 according to the European Pharmacopoeia guidelines for the production of homeopathic remedies. Texture clustering departs from obtaining a comprehensive texture representation of the full texture patch database using a convolutional neural network. This representation is then dimensionally reduced to facilitate clustering through advanced machine learning techniques. Following this methodology, 13 clusters were found and their degree of fractality determined by means of Local Connected Fractal Dimension histograms, which allowed for characterization of the different production modalities. As a consequence, each image was represented as a vector in R13, enabling classification of mixing procedures via support vectors. As a main result, our study highlights the clear differences between turbulent and laminar mixing procedures based on their fractal characteristics, while also revealing the nuanced nature of the diffusion process, which incorporates aspects from both mixing types. Furthermore, our unsupervised clustering approach offers a scalable and automated solution for analyzing the databases of evaporated droplets.

Vortex-like vs. turbulent mixing of a Viscum album preparation affects crystalline structures formed in dried droplets

“…Deep-learning based evaluation of patterns formed in dried solution droplets has proven effective in various models [19][20][21][22][23][24] and enables fast and objective image classi cation. We show here that DEM images obtained from a Viscum album quercus L. 10 − 3 solution mixed using turbulent or laminar ow show differences from a diffusion-based mixed control when subjected to semi-and fully-automated deep-learning pattern classi cation [11,12]. In our study all applied pattern evaluation approaches used (i.e.…”

“…The results of the supervised and unsupervised pattern evaluation based on deep-learning are described in detail elsewhere [11,12].…”

“…The experimental system stability was controlled by performing ve systematic control experiments following the same experimental set-up as the corresponding main experiments but using only the variant T (insigni cant results indicate a stable experimental system); the systematic control experiments gained in total 593 patterns. Additionally, the database from ve main experiments was shared with the Pattern Analysis Lab at CINVESTAV, Mexico and analyzed by supervised and unsupervised deep learning algorithms (the results of deep-leaning based pattern evaluation are published elsewhere [11,12], here we shortly present the main outcomes only).…”

“…Supervised and unsupervised deep-learning based methodologies for the evaluation of the DEM pattern database of the main experiments were developed for the purpose of this project; the protocols of these methodologies and obtained results are described in detail elsewhere [11,12]. In short: Images were subjected to a background subtraction and converted do binary.…”

Vortex-like vs. turbulent mixing of a Viscum album preparation affects crystalline structures formed in dried droplets

“…The pattern evaluation results based on deep learning deriving from the supervised and unsupervised approach are described in detail elsewhere 12 , 13 .…”