Abstract
Electron microscopy is undergoing a transition; from the traditional model of producing only a few micrographs, through the current state where many images and spectra can be digitally recorded, to a new state where a very large volume of data (movies and multi-dimensional series) can be rapidly obtained. Here we discuss the application of so-called “big-data” methods to high dimensional microscopy data, using unsupervised multivariate statistical techniques in order to explore salient image features, focusing on a specific example of domains in BiFeO3. Remarkably, k-means clustering reveals domain differentiation despite the fact that the algorithm is purely statistical in nature and did not contain any prior information regarding the material, any coexisting phases, or any differentiating structures. While this result is a somewhat trivial case, this example signifies the extraction of useful physical/structural information without any prior knowledge regarding the sample or the instrument modality. Further interpretation of such results may still require human intervention. However, the open nature of this project and the availability of software, enable broad collaborations and exploratory work necessary to enable efficient big data analysis in electron microscopy
